Home

D4.3 - Prototype Software (Documentation)

Contents

1. gt v view updates Y Settings normal speed on ticks B Epidemiology Localisation When Declared setup go 0 z E E country A a ETD United Kingdom v Addo ticks 30 om iod 2 0 r y latency pe ___ prop social media 0 70 recovery period 6 prop in target 0 10 n eel popn h 10 per 1000 popn O SD NV E I E Ave NV prop antivax 0 10 O sD Nv ph Epidemic rates 0 0050 C inc new E mo E Prev total 10 BB Protect E 0 0 m E Protect 7 haa Impact Regions Incidence N A N A N A Max When Max P When 0 0 0 0 num messages 10 Extreme Weak flu Strong flu Epidemic scenarios clear contexts Communication plans Low Medium High NV W Clear Orientation to NetLogo At this point it is worthwhile introducing some basic terminology that NetLogo uses gt Agent an entity that is able to make decisions based on its own characteristics and environment simulated people in this model gt Patch the world in the simulation is made up of a grid of squares called patches these are also referred to as regions in this simulation gt Tick each time step or advance in the simulation is called a tick 2 2 3 1 Menus NetLogo has the common menus found in most software File Edit Tools etc These allow you to open and save simulations zoom and open various NetLogo tools Note that saving the simulation saves only the model file which
2. Al Ansary L A Bawazeer G A Driel M L van Nair S Jones M A Thorning S Conly J 2011 Physical interventions to interrupt or reduce the spread of respiratory viruses Cochrane Database of Systematic Reviews 2011 CD006207 Jefferson T Di Pietrantonj C Rivetti A Bawazeer G A Al Ansary L A Ferroni E 2010 Vaccines for preventing influenza in healthy adults Cochrane Database of Systematic Reviews 2013 Keeling M J Eames K T 2005 Networks and epidemic models Journal of the Royal Society Interface 2 295 307 Kiss Z Cassell J Recker M Simon P L 2010 The impact of information transmission on epidemic outbreaks Mathematical Biosciences 225 1 10 Lam P P McGeer A 2011 Communications Strategies for the 2009 Influenza A H1N1 Pandemic National Collaborating Centre for Infectious Diseases Lin L Savoia E Agboola F Viswanath K 2014 What have we learned about communication inequalities during the H1N1 pandemic a systematic review of the literature BMC Public Health 14 484 Lorscheid Heine B Meyer M 2011 Opening the black box of simulations increased transparency and effective communication through the systematic design of experiments Computational and Mathematical Organization Theory 18 22 62 34 D4 3 Prototype Software Documentation TELL ME project GA 278723 Maurer J Uscher Pines L Harris K M 2010 Perceived seriousness of seas
3. lt B m ol v lt 2 Bo ol v v Favorites Name t Favorites Name Favorites Name 2 Dropbox Dropbox gt B array 3 Dropbox gt extrawidgets api 0 1 SNAPSHOT jar gt extensions gt bitma extrawidgets core 0 1 SNAPSHOT jar All My Files All My Files ean All My Files s sa HubNet Client 5 1 0 gt gis hamcrest core 1 1 jar iCloud Drive gt HubdNet jar 5 iCloud Drive gt E gogo 5 iCloud Drive json simple 1 1 1 jar Q AirDrop gt E lib 6 AirDrop gt matrix 6 AirDrop gt junit 4 10 jar A ie Mathematica Link A gt m network A gt widgets J Applications gt E models Applications gt Be nw 7 Applications gt xw jar E Desktop NetLogo 3D 5 1 0 C Desktop gt profiler EC Desktop Documents B NetLogo 5 1 0 M Documents a Documents NetLogo Logging 5 1 0 Downloads a NetLogo User Manual pdf O Downloads O Downloads Devices netlogo_logging xml Devices Devices Remote Disc carpe egs Remote Disc Remote Disc NetLogo 5 1 0 2 l NetLogoLite jar pack gz NetLogo 5 1 0 NetLogo 5 1 0 2 Download the TELL ME simulation zip file from either the TELL ME http www tellmeproject eu or CRESS http cress soc surrey ac uk web home website and save it on your computer Note the unzipped folder contains both the NetLogo model file TELL ME model nlogo and a folder with geographic information system GIS data These must be unzipped to the same l
4. newE 0 0314 AN LJ Inc new E a E Prev total I E E Prev total I This lower peak is because the messages were successful in encouraging vaccination previous plots and vaccination is effective in controlling the epidemic This is visible by comparing the timing with the large jump in vaccination black line occurring at the same time as the sharp break in the epidemic curves 2 5 4 Map view The map view shows the country being simulated broken into patches or regions Recall the colour of patch indicates epidemic progress red for active blue for not yet reached and green for immune or post epidemic When blue or green the darker the shade of the patch the higher the population When red the shade indicates whether the prevalence is high or low Each patch also has a small circle that indicates whether the majority of simulated people in the region have purple or have not white adopted protective behaviour For the Complete Comms Plan scenario we can see from Figure 14 that at tick time step 416 the epidemic has not spread blue over the whole of the country United Kingdom but there are small areas of green where the epidemic did infect more than half the population but has now passed 25 D4 3 Prototype Software Documentation TELL ME project GA 278723 Figure 14 Map view Complete Comms Plan a db gt ticks 416 3D 2 5 5 Monitors Recall the monitors show e Impact th
5. sliders denote the RO basic reproductive ratio latency period and recovery period of the disease being simulated The Localisation sliders give the values for proportion of the population using social media in the target group against vaccination and the number of healthcare workers per 1000 population At the bottom of the screen there is also a slider for num messages which controls the maximum number of messages available in the Communications Plan tab see section 2 4 2 2 2 5 Orientation to the TELL ME simulation outputs The main outputs of the simulation are all found in the interface tab This section explains what outputs there are with more detail on interpreting results in section 2 5 2 2 5 1 World The most immediately eye catching output is the map view Once you have clicked setup this shows the country that is currently being simulated The view is split into small squares called patches The map view is updated during the simulation and at any point of simulated time will provide broad information about the epidemic status of regions and protective behaviour adopted by simulated people within each region The colour of patches indicates epidemic progress red for active blue for not yet reached and green for post epidemic When blue or green the darker the shade of the patch the higher the population When red the shade indicates whether the prevalence is high or low Each patc
6. 01 14 2015 12 23 45 8869 0000 5 min pxcor min pycor max pycor 6 40 40 40 40 7 run number randomise country RO latency period 8 4 true United Kingdom 6 2 q 3 true United Kingdom 6 2 10 2 true United Kingdom 6 2 11 2 true United Kingdom 6 2 12 4 true United Kingdom 6 2 The first six rows describe the BehaviourSpace experiment what the file is the simulation that produced the results the experiment name the time and date and the size of the map view These are typically not of interest for analysis though may be kept for future reference Below this is the main data you are going to use to analyse the results There is a separate column for each variable input parameter or output value calculated during the run D4 3 Prototype Software Documentation TELL ME project GA 278723 The first column is a NetLogo generated run number which is different for each repetition of each parameter combination NetLogo will run simulations in parallel depending on how many processors you have SO many runs may be occurring simultaneously and will therefore have their results interleaved in the csv file This run number is followed by the columns for inputs In the screen shot you can see the run number and the first four columns of the input parameters The next column is called step and is the time step or tick If you set up the experiment to record results only at the end of each simulation there will be one row f
7. TELL ME project GA 278723 EXECUTIVE SUMMARY The agent based social simulation component of the TELL ME project WP4 developed prototype software to assist communications planners to understand the complex relationships between communication personal protective behaviour and epidemic spread Using the simulation planners can enter different potential communications plans and see their simulated effect on attitudes behaviour and the consequent effect on an influenza epidemic The model and the software to run the model are both freely available see section 2 2 1 for instructions on how to obtain the relevant files This report provides the documentation for the prototype software The major component is the user guide Section 2 This provides instructions on how to set up the software some training scenarios to become familiar with the model operation and use and details about the model controls and output The model contains many parameters Default values and their source are described at Section 3 These are unlikely to be suitable for all countries and may also need to be changed as new research is conducted Instructions for how to customise these values are also included see section 3 5 The final technical reference contains two parts The first is a guide for advanced users who wish to run multiple simulations and analyse the results section 4 1 The second is to orient programmers who wish to adapt or extend the simulati
8. allows you to run simulations If you save the simulation after a run the results are not saved If you wish to save the results you must export them File gt Export 10 D4 3 Prototype Software Documentation TELL ME project GA 278723 2 2 3 2 Simulation speed At the top of the screen you will see a slider with normal speed written below it By moving this slider you can slow down or speed up the simulation It is advised you move this all the way to the right to speed up the simulation as it will otherwise take several minutes to run However speeding up the simulation will also mean that the screen is not updated every tick 2 2 3 3 Tabs At the top of the window you will notice six tabs Interface Communications Plan Policy Context Advanced Parameters Info and Code The user can access different screens by clicking on the relevant tab All NetLogo simulations have the Interface Info and Code tabs but their content is specific to the particular model Interface displays the basic inputs controls and outputs of the simulation Communications Plan displays the set of messages and their characteristics you are about to simulate Policy Context and Advanced Parameters allow you to view and set various more advanced parameters that allow finer control of the scenario to be modelled The Info tab contains some basic info on the simulation
9. approach creates nominal compartments for people of different epidemic status such as susceptible or infected and models the rates at which people move between compartments Behaviour is added by increasing the number of compartments and adjusting rates For example the susceptible may be split into susceptible good hand hygiene and susceptible poor hand hygiene and the rate of infection for the former would be lower than the latter However these models are unable to deal with the heterogeneity requirements of TELL ME Compartment divisions would be needed not just for behaviour choices but also factors that contribute to behaviour such as different levels of perceived risk geographic proximity to the outbreak exposure to D4 3 Prototype Software Documentation TELL ME project GA 278723 communication and relevant demographic features The most appropriate modelling methodology to meet the requirements of the TELL ME prototype simulation is agent based modelling ABM because it allows heterogeneous individuals to make decisions based on their own characteristics and situation and for the actions of these individuals to change their environment Gilbert 2008 The most relevant published model is an ABM of facemask use during the 2003 SARS outbreak in Hong Kong Durham and Casman 2012 The authors present this model as the first demonstration of a quantitative HBM Health Belief Model suitable for i
10. be very time consuming to simulate The repetitions box denotes how many times each set of parameters will be run think of this as repeating the same experiment to get a sense of the variability in the simulation The next box is where you specify what outputs you wish to record in the simulation The main outputs of interest are likely to be the following enter the italic text in the Measure runs using these reporters box to include these outputs in the experiments gt Attitude to nonvaccination protective behaviour mean attitudeNV current of people gt Attitude to vaccination mean attitudeV current of people gt Proportion of population adopting nonvaccination protective behaviour count people with behave protect count people gt Proportion of population adopting vaccination count people with behave vaccinate count people D4 3 Prototype Software Documentation TELL ME project GA 278723 Y Proportion of infected adopting nonvaccination protective behaviour if count people with infected 0 count people with infected and behave protect count people with infected Incidence proportion newly infected this tick incidence Prevalence total infected proportion of the population prevalence Impact sum num infected num immune of live patches sum popn of live patches VV V Y Regions affected by the epidemic count live patches with num susceptible lt 0 95 popn count
11. not exist As more information becomes available you may also wish to permanently change the default values for some parameters so that they do not need to be altered every time you wish to run a series of simulations 2 4 1 Saving changes to parameters At this point it is important to highlight issues concerning saving changes you make to the simulation When you move a slider or selector in the main interface and save the simulation using File gt Save or Save As these changes will be saved That is the value you have selected will become the default for when you reopen the simulation another time It is therefore important not to save the model unless you intend those values to be changed Note that the original values shipped with the model can be retrieved using the default behaviour button at the bottom left of the interface However if you wish to save any of the changes you make to the Communications Plan Policy Context or Advanced Parameters tabs you must make changes directly to the code as changes in the sliders and inputs will not be saved These are available in the qui defaults procedure To find this procedure go to the Code tab and use the drop down procedures menu and select gui defaults Your cursor will move to the correct part of the code Once here you must manually change the number for the parameter s you wish to change see Section 3 5 for further explanation 2 4 2 Default
12. scenarios At the bottom right of the main interface you will six buttons that generate realistic parameter values for three epidemic scenarios and three communications plan scenarios By clicking these buttons the simulation automatically sets the parameters to the appropriate values These scenarios may be helpful as a starting point in helping you decide what parameter values you would like to explore The epidemic parameters are based on World Health Organisation and United States United States Centers for Disease Control and Prevention planning assumptions and the communications plan parameters are based on 18 D4 3 Prototype Software Documentation TELL ME project GA 278723 government reports and discussions with communications experts from the United Kingdom Table 2 details what the scenario buttons do Table 2 Scenario buttons Scenario Action Button Changes the Epidemic Parameters to RO 1 4 latency period 2 recovery period 5 Does not change any other parameters Changes the Epidemic Parameters to RO 2 latency period 1 recovery period 5 Does not change any other parameters Changes the Epidemic Parameters to RO 6 latency period 7 recovery period 21 Does not change any other parameters Changes the communications plan to a low activity plan View the Communications Plan tab to view the settings for the messages Changes the communications plan to a medium activity plan View th
13. select which country you would like to simulate Note that selecting a country does not automatically load it we have to press setup for this to happen 2 2 4 3 Sliders Beneath the three red headings Behaviour Epidemiology and Localisation you will notice eleven sliders These sliders are used to tell NetLlogo what numbers you want to use for the parameters named on the sliders You can change the value simply by clicking and holding on the coral coloured marker and moving it to the value you wish or by clicking on the slider image next to the marker which will move it up or down incrementally The value of the slider s parameter or where the marker is located is displayed in the bottom right corner of the slider 11 D4 3 Prototype Software Documentation TELL ME project GA 278723 The model simulates two types of behaviour vaccination V and other protective nonvaccination behaviours such as hand washing NV The four Behaviour sliders denote the efficacy and threshold for each of these behaviours Efficacy is the reduction in epidemic infectivity if the behaviour is adopted so a higher value will result in a smaller epidemic if people adopt the behaviour The threshold is the behaviour score which must be reached for a person to adopt the behaviour so a higher value will make it harder to reach the required score and lead to a smaller number of people adopting the behaviour The Epidemiology
14. stays there until it drops to zero at the end of the epidemic this occurs later in the more complete communications plan scenario The proportion of the total population adopting nonvaccination protective behaviours blue follows a similar path though with less noise and does not drop off to zero at the end showing these behaviours have endured post epidemic Finally the black line showing the proportion of the population that adopts vaccination shows clearly different results between the two scenarios In the complete communications plan scenario we see a jump after the start of the epidemic but in the one message scenario we see vaccination rates barely rising above zero 2 5 3 Epidemic rate plot The epidemic rates plot shows the incidence proportion newly infected people this time step in pink and the prevalence total proportion infected in red We can see in Figure 13 and Figure 12 that the two scenarios give very different curves suggesting the different communications plans have had differing effects on the epidemic The more complete plan scenario shows an overall lower level of incidence and prevalence Note that the NetLogo software automatically scales the plots so the left hand plot has a much lower peak prevalence about 1 compared to 3 even though the peaks are of similar heights Figure 13 Epidemic Plot Complete Comms Plan Figure 12 Epidemic Plot One Msg Comms Plan Epidemic rates Epidemic rates 0 0137 Llinc
15. 10 million simply enter 10 To find this procedure open the code tab and then use the procedures drop down box The code should be inserted in addition to the existing code and be in the same format You can check that NetLogo has accepted the syntax of the code by clicking on the green tick at the top left of the screen Figure 19 New code for new country 1f country lt new countrys set popn dataset gis load dataset GISdata Popn density lt new country 2015 asc set total popn 777 The final step is to provide the GIS data for NetLogo to load The data was taken from the SEDAC website Center for International Earth Science Information Network CIESIN Columbia University and Centro Internacional de Agricultura Tropical CIAT 2013 http sedac ciesin columbia edu data set gpw v3 population density future estimates which has data available for all countries You should find the gridded projected population density and download the 2015 data in ascii format Once you have the file you should place it in the GIS folder that you downloaded with the simulation and name it in using the same 32 D4 3 Prototype Software Documentation TELL ME project GA 278723 convention as the other files present i e Popn density new country 2015 asc If necessary trim the data to exclude islands or other distant features using GIS software such as the open source Quantum GIS so that only the regions
16. 2014 By failing to prepare you are preparing to fail lessons from the 2009 H1N1 swine flu pandemic The European Journal of Public Health Department of Health UK 2012 UK Pandemic Influenza Communications Strategy 2012 Department of Health UK Durham D P Casman E A 2012 Incorporating individual health protective decisions into disease transmission models a mathematical framework Journal of The Royal Society Interface 9 562 570 Eurostat 2014 Households level of internet access Eurostat Ferrante G Baldissera S Moghadam P F Carrozzi G Trinito M O Salmaso S 2011 Surveillance of perceptions knowledge attitudes and behaviors of the Italian adult population 18 69 years during the 2009 2010 A H1N1 influenza pandemic European journal of epidemiology 26 211 219 Funk S Gilad E Watkins C Jansen V A A 2009 The spread of awareness and its impact on epidemic outbreaks Proceedings of the National Academy of Sciences 106 6872 6877 Funk S Salath M Jansen V A A 2010 Modelling the influence of human behaviour on the spread of infectious diseases a review Journal of The Royal Society Interface 7 1247 1256 Gilbert N 2008 Agent Based Models SAGE Publications London United Kingdom International Labour Organisation 2014 SEGREGAT Employment for detailed occupational groups by sex International Labour Organisation Jefferson T Del Mar C B Dooley L Ferroni E
17. SEVENTH FRAMEWORK PROGRAMME saat oe E sS D4 3 Prototype Software Documentation 2nd Reporting period WP4 Agent Based Social Simulation Responsible Partner SURREY Due date of the deliverable M36 January 31st 2015 Actual submission date M36 January 31st 2015 Dissemination level PU TELL ME Transparent communication in Epidemics Learning Lessons from experience delivering effective Messages providing Evidence Project co funded by the European Commission within the 7 Framework Programme HEALTH theme Document Management PROJECT FULL TITLE PROJECT ACRONYM GRANT AGREEMENT STARTING DATE DURATION D4 3 Prototype Software Task 4 3 D4 3 Prototype Software Documentation TELL ME project GA 278723 Transparent communication in Epidemics Learning Lessons from experience delivering effective Messages providing Evidence TELL ME Collaborative Project funded under Theme HEALTH 2011 2 3 3 3 Development of an evidence based behavioural and communication package to respond to major epidemics outbreaks 278723 01 02 2012 36 months Leader SURREY Other contributors none D4 3 Prototype Software Documentation TELL ME project GA 278723 Table of Contents EXECUTIVE UMMAR eere E E 4 De MN o D t S1 i PE EEE E E cen perro acces sn E E eae Deere E E E E E aces 5 1 1 Advancing modelling of behaviour during epidemics cccccssseccceeseccceesecceeeese
18. a new country The prototype has been set up to simulate 28 European countries It uses GIS data to map each country and its population density across the different regions patches One of most straightforward changes you may wish to make is to run a communication and epidemic simulation for a country that is not already included We will use a walk through of how to do this to introduce you to some key concepts in changing the simulation The NetLogo platform automatically connects interface controls to programming code The first step is to add the new country to the drop down box on the interface as in Figure 18 Right click on the drop down box and click edit then type in the name of the country to be added Click OK and then check that the country appears in the dropdown list Figure 18 New Country Drop down Box K ESR Chooser Global variable country Choices New Country 0 Austria Belgium Bulgaria Croatia Cyprus Czech Renublic example a b c 3 45 Now we need to change the NetLogo code to include the instructions to load the map and population density data for the new country Figure 18 shows a template for the new code which should be added inside the load country data procedure There are two components the filename for the GIS data and the population value The total popn value is the estimate population in 2015 and should be entered in millions e g for a country with a population of
19. alues so there has been no attempt to provide locally appropriate default parameter values The only localisation included in the simulation is the GIS information that provides a map and population density for each country Instead a single data source was used where available to set a reasonable value The default values for parameters set with sliders or other controls on the main interface can be easily changed The current value is saved with the simulation However parameter values in the other interface screens will revert when the simulation is reopened and their defaults must be changed by editing the model code see section 3 5 3 1 Epidemic parameters Users can easily change the parameters related to the underlying nature of the epidemic but are three predefined scenarios in the simulation see Table 5 that can be retrieved by clicking the relevant button Default behaviour efficacy values are set at 0 7 for vaccination Jefferson et al 2010 Valenciano et al 2011 and 0 25 for nonvaccination Jefferson et al 2011 Table 5 Epidemic scenarios data Values Data sources Weak flu RO 1 4 latency period 2 WHO Pandemic flu guidelines WHO 2013 recovery period 5 Strong flu RO 2 latency period 1 WHO Pandemic flu guidelines WHO 2013 recovery period 5 Extreme RO 6 latency period 7 Based on Rubella Centers for Disease Control and Prevention recovery period 20 2012 3 2 Communications
20. ameters on the main interface If you wish to vary Communications Plans Policy Context and or Advanced Parameters you will need to consult the documentation at https github com CRESS Surrey extraWidgets wiki Using the extension with BehaviorSpace This is only suitable for advanced users with NetLogo coding skills The basic premise of BehaviorSpace is that it performs multiple runs of the simulation with different parameter values and thus scenarios chosen by the user records the outputs of the simulation and saves all of this in a csv comma separated values formatted file These are then available for further analysis with spreadsheet e g Excel or statistical e g R software This provides a way to systematically examine the effect of different parameter values on the simulation behaviour Lorscheid et al 2011 Railsback and Grimm 2012 4 1 1 Creating an experiment To use BehaviorSpace select Tools gt BehaviorSpace and select New This opens a new experiment window You can give the experiment a name at the top of the window In the second box you can specify which parameters you wish to vary NetLogo automatically lists all of the parameters on the interface you just need to list or give a start increment and end point to specify the values you wish to run see notes underneath the box BehaviorSpace will run every combination of values you specify so beware of creating a very large number of values as this will
21. ancing modelling of behaviour during epidemics The TELL ME project relies on the connection between protective behaviour and epidemic transmission That is personal voluntary decisions to be vaccinated or adopt hand hygiene and social distancing measures are expected to reduce the impact of an influenza epidemic Without such a connection there would be no value in communication encouraging such behaviour However these decisions are influenced by the person s situation which includes proximity to and other features of the epidemic The connection between communication and behaviour should be consistent with both the relevant psychological theories and any empirical data In order to provide a realistic simulation for planning purposes there are five requirements for the model design gt Communication affects behaviour gt Behaviour is based on appropriate psychological models gt Heterogeneity of behaviour response and situational awareness gt Two way dynamic influence of epidemic on behaviour and behaviour on epidemic and gt Parameterisation with empirical data where available A recent review Funk et al 2010 identified 25 theoretical studies that considered the mutual influences of personal behaviour and epidemiology of infectious disease the fourth requirement Almost all the models reviewed were compartment models which is the standard mathematical approach to modelling epidemics using differential equations This
22. associated with the effect of government communication on an individual s behaviour are a prime example Discussion and training tool it is most likely that all of the above uses will be most beneficial when conducted with colleagues and other stakeholders In this group format the simulation becomes a discussion tool which can help identify areas of consensus and disagreement and develop a shared understanding of the effect of communication and improve decision making capability 2 User Guide This guide is intended for all users of the TELL ME simulation regardless of prior experience with using models No programming is required to use the simulation it is operated simply by entering parameter values and pressing buttons The user guide contains three parts gt a quick start guide which describes installation of the required software and an orientation to the interface including a description of the key input parameters and simulation output gt a guided set of scenarios to assist users to learn how to input their own scenarios and gt adetailed examination of simulation outputs 2 1 Modelling environment The simulation is developed in NetLogo Wilensky 1999 a specialist ABM software application with its own programming language The TELL ME simulation enables users to input communication plans and also to manipulate other parameters that are relevant for planning such as the country to be considered and the infectivity o
23. ation or behaviour adoption 2 Ineffective With the same baseline epidemic behaviour is added to the simulation with low behaviour thresholds but is ineffective 3 Effective Next the efficacy of behaviour is raised and adoption of behaviour therefore has an effect on the epidemic 4 Unresponsive This scenario has higher behaviour thresholds making adoption less likely to provide a comparison base for the communications scenarios 5 Basic Communication This scenario introduces simple communication Regular messages about benefits of vaccination attitude plus once off emphasis of responsibility norms 6 Communication Plan The final scenario has a combination of messages with different media and 2 3 1 effects Scenario 1 Epidemic curve The purpose of this scenario is to show the way in which the epidemic spreads in the simulation if nobody adopts protective behaviour It uses default values for all parameters except that the behaviour thresholds are set to 1 so that behaviour is never adopted and increases RO so that an epidemic happens more quickly There is no communication plan Setup instructions gt VV v gt Open or close without saving then reopen the simulation to remove any changes you may have previously made Choose any country Set up the infection to have an RO of 3 by moving the RO slider to 3 Set the protectV threshold slider to 1 se
24. ceceeseceseeeceeeeueeeetes 5 1 2 POST Oy SINC reen e E E E E E 6 1 3 How we envisage you will use the simulation ssessssssseessssersssrrressreresrrressrerssrrressreresrrreserereserresseeee 7 NGS serere E E E E 8 2 1 Modene CnN eere EEA E EEE 8 2 2 E S EE r E ee re eee ee ears 9 2 3 Learning to use the simulatio Meseria E 13 2 4 Scenario management ss ssessessessessessrsseesersesssesessersersrrsrrsresresrrsresresresrroseoreoreoseoseoseesessesseeseesees 18 2 5 Understanding the results Of the SIMULAtTION ccccccccssseccceesececeeeceeeeeseceeseececseaeceeseneceeseneeetes 23 3 Detauit Vases Or Pal aie ve a N E E E EE E EE 27 3 1 Epidemic parame tE ersan EEE 27 3 2 COMMUNICATIONS plan ScCENariO S sesscicicesssanesiosiasnosaessdnowas uaccboasavanienia A E E 27 3 3 Population structure parameters essssssssesessresesreresrrersrrrresererssrrressrerssrtresereessetreserereseeresereessreressre 28 3 4 B haviour decision paralinet eT Sisirin innana ANN 28 3 5 Customising default parameter ValUCS c cccsecccccssecccceesececeesececceecceceuecceeeeneceseegeceeseuecessegeceeseges 28 A TECNICA RETT OEE e T E 29 4 1 Running your own systematic experiments ssessessessessersersrrsrrsrrsrrsresrrsresresresresseeseeseeseesesseesese 29 4 2 Changing the simulaatioissa se natesesetincunaveicoenerencesacuioeanect same sewuuecdecs 31 SE OC ya ere ee eee Te err ee ee ee ee ee ee 34 D4 3 Prototype Software Documentation
25. e Communications Plan tab to view the settings for the messages Changes the communications plan to a high activity plan NB this plan has no High NV vaccination View the Communications Plan tab to view the settings for the messages There are also three buttons at the bottom left of the interface default behaviour resets the behaviour parameters to those when the simulation is first opened The clear messages and clear contexts buttons clear the Communications Plan and Policy Context tabs respectively 2 4 3 Communications plan Comparing the effect of different communications plans is the main function of the TELL ME simulation Therefore communications plans are likely to be one of the key scenario elements you would like to change From using the learning scenarios you should now understand how to change communication plans so we shall now explain how communications plans are defined in the simulation A plan is made up of a group of messages Each message has six parameters and occupies one row in the Communications Plan screen The messages that make up a plan are described in the simulation using the six parameters detailed in Table 3 Table 3 Communications plan message parameters Parameter Values Description Trigger Before This parameter in combination with the Day Level below Start identifies when a message will occur or be triggered Messages Day can begin i before the start of an e
26. e any protection and therefore has no impact on the epidemic spread 2 3 3 Scenario 3 Effective The purpose of the third scenario is to isolate the effect of protective behaviour on the spread of an epidemic The same settings will be used as in the previous scenario except that the efficacy of the behaviour is more realistic returned to their default values As before the randomise checkbox is unchecked so the same epidemic is being run Setup instructions gt Set efficacy vaccine to 0 7 gt Set efficacy protect to 0 25 Run the simulation remember to press setup first The epidemic should be both smaller lower impact value lower maximum incidence and maximum prevalence and later higher values for when than in the previous scenario That is the protective behaviour is helping to control the epidemic In addition you may be able to observe that the maximum behaviour adoption high points on the blue and black curves are lower than in the previous scenario This is because a smaller epidemic also reduces the incentive to adopt protective behaviour This result demonstrates the importance of modelling the two way feedback between the epidemic and the behaviour and the difficult of understanding the complexity of the feedback without a simulation 2 3 4 Scenario 4 Unresponsive The purpose of the fourth scenario is to set up some space to see the effect of communication Behaviour threshold
27. e proportion of the population that are or have been infected during the epidemic e Regions the proportion of regions patches in the simulation in which more than 5 of the population have been or are infected during the epidemic e Incidence the total number of people who have become infected during the current time step tick e Max the highest incidence during the current simulation proportion of population e When during which time step tick the highest incidence occurred e Max P the highest prevalence during the current simulation and e When during which time step tick the highest prevalence occurred Figure 15 Monitors Complete Comms Plan Figure 16 Monitors One Msg Comms Plan Impact Regions Incidence Impact Regions Incidence 0 27 0 7 589 0 5 0 81 1 Max When Max P When Max When Max P When 0 0024 102 0 0127 110 0 0051 149 0 0301 156 We can see from Figure 15 and Figure 16 that the impact and region monitors were higher in the one message plan as were the Max I and Max P That is the complete communications plan was more effective in controlling the epidemic by encouraging protective behaviour 26 D4 3 Prototype Software Documentation TELL ME project GA 278723 3 Default Values for Parameters It is likely that the real value for parameters in the model is country or even region specific The simulation is intended as a prototype and there is limited availability of data to set these v
28. ealth care workers iv people currently infected or v people with anti vaccination views This parameter denotes the medium through which messages are delivered to the target group They can be channelled via i the mass media ii social media iii the health media to health care workers or iv by health professionals to the public This parameter describes the content of the message This can be i to provide information on the current state of the epidemic which increases trust ii to explain the benefits of adopting a protective behaviour which increases attitudes iii to emphasise that people have a responsibility to help combat the epidemic by changing their behaviour which temporarily adds a bonus to the behaviour score or iv to raise awareness of the risk of the epidemic which temporarily makes people reassess the threat they perceive as if they were close to the epidemic front This parameter identifies which behaviour type is the subject of the message This can be i related to vaccination only ii related to other protective behaviours only e g handwashing or iii related to both of the above The simulation will not check that you have created a sensible message or plan for example you could create a message intended for everyone target of All but delivered in specialist health publications media channel of Health media You should take the time to verify that each message a
29. ealthcare O 1 The amount that a person s attitude will change if they receive advisor information from a healthcare worker Trust bonus from 0 0 2 Increase in trust due to provision of information about epidemic information status Duration of norms 0 30 days The number of days for heightened awareness of norms or bonus recommendation after message emphasising responsibility Bonus for vaccination 0 0 5 Additional norms score above the proportion of nearby people who have adopted behaviour for vaccination Bonus for other Additional norms score above the proportion of nearby people protective behaviour who have adopted behaviour for nonvaccination protective behaviour Daily discount for past 0 0 2 The discount per time step for epidemic history in risk calculations incidence Threat perception 1 5 Multiplier for threat perception compared to H1N1 multiplier Vaccination attitude 0 1 Weight of own attitude in seeking vaccination weight Vaccination norms Weight of norms proportion of nearby people who have adopted weight behaviour in seeking vaccination behaviour calculation Nonvaccination Weight of own attitude in adopting nonvaccination protective attitude weight behaviours Nonvaccination Weight of norms proportion of nearby people who have adopted norms weight behaviour in calculation for other protective behaviour 2 5 Understanding the results of the simulation Now we have described all of the paramete
30. eters The Advanced Parameters tab contains a further set of parameters used in the simulation These are set using real world data where possible They can be changed but it is envisaged only advanced users will do this Table 4 describes the available parameters and their interpretation More details are available in the design documents particularly TELL ME Report D4 2 Table 4 Advanced parameters Parameter Values Description Travel rate 0 1 The proportion of an individual s contacts outside of their region patch Travel proportion The proportion of an individual s contacts outside of their short region patch that are with the neighbouring regions patches Prevalence at which User number input The national prevalence at which the simulation considers the epidemic declared epidemic to have started See distance Radius for scope of vision for individuals to perceive whether agents have adopted protective behaviour and the number of new infections in other regions patches Population 1000 2000 5000 The number of people represented by one agent in the simulation represented by or 10000 person Randomise ON or OFF Turns on or off the random number generator this should be left on unless you wish to reproduce the exact same epidemic and behaviours for comparison purposes Random seed value User input Reports the current random seed value and can be used if monitor Randomise Is off to enter a previously used rando
31. f the disease A basic NetLogo simulation has three components The Interface component provides tools to allow the user to manipulate key parameters and run the simulation and charts and other information to monitor the simulation during the run The Info component is to allow accessible documentation to be packaged with the simulation this gives only very basic information The Code component sets out agent properties interaction rules and data That is this provides the programming to implement the simulation While the code is available to the model user the user does not need to do any coding The TELL ME simulation uses an extended version of NetLogo to provide additional Interface components This means that the main Interface component contains only the most important variables that a user may wish to access The additional interfaces allow more detailed control of communications plans epidemic management policies and parameters that are internal to the simulation As well as providing the structure and tools for building an agent based model NetLogo also provides tools for analysing the results of simulations For example the BehaviorSpace tool provides scenario management capabilities so that results from multiple simulation runs can be exported for analysis D4 3 Prototype Software Documentation TELL ME project GA 278723 2 2 Quick start guide 2 2 1 Installation Installing the TELL ME sim
32. fluenza Monitoring Vaccine Effectiveness in Europe I MOVE multicentre case control study PLoS medicine 8 e1000388 Vaux S Van Cauteren D Guthmann J P Le Strat Y Vaillant V Valk H de L vy Bruhl D 2011 Influenza vaccination coverage against seasonal and pandemic influenza and their determinants in France a cross sectional survey BMC Public Health 11 30 Vidal J 2009 Fundamentals of Multiagent Systems with NetLogo Examples Scribd We Are Social 2014 Social Digital amp Mobile Around The World We Are Social WHO 2013 Pandemic Influenza Risk Management WHO Interim Guidance World Health Organization Wilensky U 1999 NetLogo 35
33. h also has a small circle which indicates whether the majority of simulated people in the region have purple or have not white adopted protective behaviour 2 2 5 2 Plots Next there are the Attitude and Behaviour plots on the left of the screen underneath the Behaviour sliders split by Vaccination V and other protective behaviours such as hand washing or wearing masks NV The attitude for each agent simulated person is scored between O and 1 for each behaviour type and the display shows the average attitude score over all agents towards the categories of behaviour The behaviour plot shows the proportion of the population who have adopted each behaviour with an extra plot in red for the proportion of those Exposed and Infected that have adopted other protective behaviours To the right there is the Epidemic Rates plot which shows the incidence and prevalence of the epidemic in each time step All three plots show the outputs over time you will see new points plotted as the simulation runs 12 D4 3 Prototype Software Documentation TELL ME project GA 278723 2 2 5 3 Monitors Below the Epidemic rates plot are seven monitors Monitors display the value of some calculation The epidemic monitors show gt Impact the proportion of the population that are or have been infected during the epidemic gt Regions the proportion of regions patches in the simulation in which mo
34. h parameter has a line in the code that is in the form set lt name gt lt value gt where lt name gt is the name of the parameter and lt value gt is the default value that is provided for the control such as a slider when the simulation is opened Identify the parameter whose value you wish to change and edit the number to your preferred value gt Save the simulation Close it and then open it Check if the parameter is now set to the appropriate value 28 D4 3 Prototype Software Documentation TELL ME project GA 278723 4 Technical Reference This section of the documentation is not required for general users of the TELL simulation model Instead it is intended for advanced users who wish to use the simulation scientifically or for programmers who wish to modify the model The reference will gt describe how to set up your own experiments and access the results gt introduce the general approach to changing the code behind the simulation and gt show a simple example of modifying the model adding another country to be simulated 4 1 Running your own systematic experiments The NetLogo platform includes a tool called BehaviorSpace which allows users to run multiple simulations This manual provides a general introduction with detailed information available from the NetLogo website at http ccl northwestern edu netlogo docs use the link on the left to BehaviorSpace BehaviorSpace can only be used to vary par
35. ible Setup instructions gt Clear existing message by pressing the clear messages button bottom left of main interface gt Create the Communications Plan in Figure 4 gt Exaggerate the effect of communication in the Advanced parameters screen bonus for vaccination and bonus for other protective behaviour to 0 5 triple all the values for the four Change parameters eg Change mass media from its default of 0 05 to a value of 0 15 Run the scenario There should be higher levels of behaviour adoption and therefore a lower epidemic impact than in the previous scenario You have now completed the teaching scenarios intended to prepare you to run your own experiments You should be gt familiar with how to set up and run the simulation model gt able to change many of the parameters in the simulation and gt understand and explain the main outputs of the simulation Close the simulation without saving and reopen it to return parameters to their default values 17 D4 3 Prototype Software Documentation TELL ME project GA 278723 Figure 4 Scenario 6 Communications Plan Frese m1 Day Level rm group m1 Media channel M Content m1 Behaviour type m1 Regular 10 Mass Mass media Promote Benefits Bi Both B Trigger m2 Day Level rem group m2 Media channel SET Content TT Behaviour Tee m2 National Prevalence 0 01 Social media media Emphasise Responsib
36. ility Responsibility Both a Trigger m3 Day Level Target group m3 Media channel m3 Content m3 Behaviour type m3 Start 34 0 Health workers Le Health media ie Promote Benefits 3 Both HA Trigger m4 Day Level Target group m4 Media channel m4 Content m4 Behaviour type m4 Regular B 25 Anti vaccination Bi Mass media B Promote Benefits Bi Vaccination B Trigger m5 Day Level i group m5 Media channel oTa Content meree Behaviour oaacmsac ie m5 _ National Prevalence 0 0080 Infected Health Health profession Promote Promote Benefits l Other Protective Protective Trigger m6 Day Level Target group m6 Media channel m6 Content m6 Behaviour type m6 National Prevalence 0 025 l All Le Mass media Le Recommend Adoption LHI Other Protective 34 2 4 Scenario management In the learning section we gave you changes to make to various parameters to create learning scenarios It is important now to understand how and why you may make changes to parameters in the simulation yourself to create the scenarios you would like to simulate This section explain what parameters are available what they control and how to change them Default values for many of these parameters are based on real world data see section 3 However there are many others that are somewhat arbitrary particularly those concerning the effect of communication because real world data do
37. in text form and the Code tab contains the simulation code that is the program that runs the model 2 2 4 Orientation to the TELL ME model input controls The simulation can be run entirely from the Interface screen The controls and other elements on this screen are described below 2 2 4 1 Buttons The most basic controls are buttons In the TELL ME simulation the two most important are setup and go found in the top of the interface just to the right of centre The setup button initialises the simulation scenario and the go button runs the simulation Other buttons at the bottom of the interface are used to clear changes that have been made by the user or to select predefined scenarios described in section 2 4 A button on the interface is associated with a section of code in the simulation code tab By pressing the button we are instructing NetLogo to run that code Buttons can either be a once button or a forever button Once buttons run that code once and then stop Forever buttons run that code repeatedly forever until some condition is met which tells the simulation to stop or you press it again The setup button is a once button and the go button is a forever button 2 2 4 2 Drop down boxes There is only one drop down box on the interface the country selector found underneath the Localisation heading By clicking on the icon we are able to
38. ing characteristics of the disease that you may wish to change see Figure 6 They are found on the main interface Figure 6 Epidemiology parameters gt RO the basic reproduction ratio this can be thought of as the number of cases one case generates on average over the course of its infectious period gt latency period the average number of days after exposure before people are infectious gt recovery period the average number of days that people remain infectious 2 4 6 Localisation parameters There are five basic localisation parameters that you may wish to change see Figure 7 They are Figure 7 Localisation parameters gt country to select the country that is simulated gt prop social media which gives the proportion of the population that have access to and use social media gt prop in target which gives the proportion of the population that receive messages that are in the high risk group used in messages e g elderly poor health status gt popn hcw which gives the proportion of the population that are health care workers gt prop antivax which gives the proportion of the population that have k b strong anti vaccination views 2 4 7 Policy context parameters The Policy Context screen contains a set of parameters that provide further situational characteristics for the simulation These do not relate directly to the behaviour of individuals or the disease characteri
39. live patches Max l max incidence When when max incidence VV Y Max P max prevalence gt When when max prevalence If you wish you can also export the map view as an image file at the end of each run by adding the following code to the final commands box on the experiment window export view word view_from_run_number behaviorspace run number png Next you select whether you wish to measure these outputs at every time step or just at the end of the simulation You can normally leave the remaining options as they are Press OK Now select your experiment and hit RUN At the next window select Table output only and hit OK Next choose where to save the csv file Finally you can speed up the experiments by moving the speed slider to the right and unchecking the two updates boxes 4 1 2 Analysis BehaviorSpace generates a results file in csv Comma separated values format Part of an example is shown at Figure 17 These files may be very large For example an experiment that tests the effect of four inputs with three values each and 20 repetitions of each parameter combination would have 1 620 separate runs 3x3x3x3x20 If each of these has 200 time steps and you set up the BehaviorSpace to record values each time step there would be 324 000 lines of output Figure 17 Example BehaviorSpace csv Output A B C D E 4 BehaviorSpace results NetLogo 5 1 0 Lt 2 TELL ME model nlogo 3 curve max 4
40. m seed value Initial attitude bonus 0 0 3 The difference in initial attitude between those in and out of the for target group target group Daily return to initial 0 10 The percent decay per time step to return to initial attitude if no attitude further communication See mass media The probability of a person receiving a mass media message See social media The probability of a person with access to social media receiving a social media message 22 D4 3 Prototype Software Documentation TELL ME project GA 278723 Parameter Values Description Healthcare worker 0 1 The probability of healthcare worker responding to messages sees health media delivered through the health media Go to doctor during 0 1 The probability of a member of the public receiving a message campaign delivered by healthcare workers Latitude of 0 0 5 The distance from their own attitude that a message will not be acceptance immediately ignored Effectiveness of 0 1 The relative effect of repeated messages to allow later messages repeat message to have a lesser effect than earlier messages Change mass media 0 1 The amount that a person s attitude will change due to mass media messages Change socialmedia 0 1 The amount that a person s attitude will change due to social media messages Change media for 0 1 The amount that a healthcare worker s attitude will change due to healthcare workers messages in specialist health media Change h
41. manently by saving the simulation with the sliders set to the desired default values 3 4 Behaviour decision parameters A calibration process was used to choose appropriate values for the behaviour parameter values based on behaviour during the 2009 H1N1 epidemic These parameters are the weights for different components attitude proportion already adopted the discount that applies to earlier cases in assessing threat and the threshold that must be met for the behaviour to be adopted The calibration used Hong Kong hand washing data Cowling et al 2010 for nonvaccination and French Vaux et al 2011 and Italian Ferrante et al 2011 vaccination data 3 5 Customising default parameter values The parameter values set by sliders and other controls on the main interface are saved with the simulation That is if you wish to use different default values simply change to the values you prefer and save the model The controls will be set to those values when you open it again However if you wish to change the values that are automatically set in the Advanced Parameters screen or the Policy Context when the simulation is opened you will need to make changes in the NetLogo code To do this gt Click on the Code tab to switch to the code screen gt Click on the Procedures dropdown box and select gui defaults from the list of procedures This will move the cursor to the correct part of the code gt Eac
42. n different scenarios and compare the outputs Once you understand the different scenarios that can be simulated you may wish to run those you consider to be of interest and explore the difference in outcomes The most obvious scenarios are the differing communication plans however you may also wish to consider different individual behaviour epidemic parameters or policy contexts Formal thought experiment the simulation is based on existing theory from the psychology and health literatures It therefore represents a formal thought experiment on the implications of these theories and allows users to track the effects and outcomes of the various theoretical assumptions If you find results in the simulation you disagree with it could be useful to explore what is causing the result and seeking to understand what specific effect it is that produces this result and thus consider what it is that you disagree or agree with from the literature or simulation Informing data collection the simulation has many parameters that are necessary to connect the theoretical understanding into practical output While some of these parameters are calibrated with real world data there are many others for which data were not available The simulation can help identify what D4 3 Prototype Software Documentation TELL ME project GA 278723 parameters strongly affect the results and would therefore be useful to collect data about in the future The parameters
43. ncorporation in agent based epidemic simulation Durham and Casman 2012 The modelled facemask adoption occurs in response to dynamic epidemic information about prevalence and deaths but there is no influence of facemask use on the progress of the epidemic The authors explicit recognise the model s limitations but also the need to compromise realism to achieve model feasibility The TELL ME model is ambitious advancing modelling about health behaviour during influenza epidemics in several ways simultaneously By including communication and mutual influence however abstractly the TELL ME model is a substantial step to extend and generalise modelling of protective behaviour and epidemic progress 1 2 Model overview The TELL ME model is a two layer simulation One layer consists of simulated individuals that receive communication messages adjust their attitudes accordingly perceive their situation and make decisions about whether to adopt or cease protective behaviour This behaviour is founded on a hybrid psychological model that includes attitudes subjective norms and perceived threat Each simulated person calculates a weighted average of their own attitude a value between O and1 the proportion of nearby people who have adopted the behaviour a proxy for norms and a threat score which is high near the epidemic frontier and lower elsewhere That risk score is the discounted cumulative local incidence that is it increases as new case
44. nd the overall plan you create make conceptual sense You can set the number of messages by choosing how many are active a message is inactive when the trigger is set to NONE If you wish to have more than 10 messages use the num messages slider on the interface you will need to press clear messages and then setup to create the extra message slots As this will also clear whatever message information you have already entered it should be done before defining the communications plan 2 4 4 Behaviour parameters There are four basic behaviour parameters parameters associated with individuals protective behaviour that you may wish to change see Figure 5 These were used during the learning scenarios they are Figure 5 Behaviour parameters gt efficacy protect this gives the proportional reduction in risk to self and others if non vaccination behaviour adopted gt protectNV threshold this gives the value for attitude norms and incidence combination required to induce protective behaviour gt efficacy vaccine this gives the proportional reduction in risk to self and others if vaccinated gt protectV threshold this gives the value for attitude norms and incidence combination at which agents seek vaccination 20 D4 3 Prototype Software Documentation TELL ME project GA 278723 2 4 5 Epidemiology parameters There are three basic epidemiology parameters that determine the underly
45. ocation with the GIS data in a folder name G Sdata and that folder in the same folder as the model file You can now open the model file either by double clicking it you may have to tell your operating system that NetLogo is the program to open this type of file or by using File gt Open from within NetLogo itself 2 2 2 Basic simulation control Once you have opened the simulation you should see a window like that shown in Figure 3 This shows the main interface of the simulation The screen has several elements to control the simulation and display the output In addition there are general controls for the NetLogo software which are the same for any NetLogo model In this quick start D4 3 Prototype Software Documentation TELL ME project GA 278723 section we will introduce the basic simulation controls A more comprehensive description of how to use the simulation is presented in Section 2 3 Figure 3 Initial simulation view F Add Button v Behaviour E eed efficacy protect 0 25 mE E protectNV threshold 0 40 E bel efficacy vaccine 0 90 eee bee protectV threshold 0 45 Attitude Score Behaviour Adopting oO NetLogo TELL ME model Users pj0017 Dropbox TellMe Model Code Bits Interface Communications Plan Policy Context Advanced Parameters Info Code default behaviour Command Center observer gt 2 2 3 clear messages
46. on model section 4 2 This material is not suitable for general users D4 3 Prototype Software Documentation TELL ME project GA 278723 1 Introduction The TELL ME project developed an evidence based communication package to respond to major infectious disease outbreaks notably influenza epidemics The project was intended to assist health agencies to develop plans to communicate before during and after any outbreak in an effective way so as to encourage appropriate population behaviour and minimise the impact of an epidemic One of the tools developed is prototype software an agent based social simulation to assist with planning decisions about communication The simulation is intended to provide decision support for health agencies and other official information providers assisting relevant officials to understand the complex problem of communicating effectively before during and after an influenza epidemic More specifically it is to allow comparison of options for communication plans with the user to enter a communication plan and explore some of the key effects on behaviour and consequently on the progress of the epidemic see section 1 3 for some potential model uses The basic question for the model is given a specific communication plan and epidemic parameters gt What proportion of the population adopt protective behaviour gt What proportion of the population is infected over the duration of the epidemic 1 1 Adv
47. onal and A H1N1 influenzas attitudes toward vaccination and vaccine uptake among US adults does the source of information matter Preventive medicine 51 185 187 ME T 2012 D1 1 Systematic Review Report TELL ME Project National Centre for Social Research 2012 National Travel Survey 2002 2012 Department for Transport Odone A Ferrari A Spagnoli F Visciarelli S Shefer A Pasquarella C Signorelli C 2014 Effectiveness of interventions that apply new media to improve vaccine uptake and vaccine coverage Human Vaccines amp Immunotherapeutics 11 e34313 Perra N Balcan D Goncalves B Vespignani A 2011 Towards a Characterization of Behavior Disease Models PLoS ONE 6 e23084 Railsback S F Grimm V 2012 Agent based and individual based modeling a practical introduction Princeton University Press Rubin G Potts H Michie S 2010 The impact of communications about swine flu influenza A H1N1v on public responses to the outbreak results from 36 national telephone surveys in the UK Health Technology Assessment 14 183 266 Teahan W 2010 Artificial Intelligence Agents and Environments Bookboon Trading Economics 2015 Population Data Trading Economics Valenciano M Kissling E Cohen J M Oroszi B Barret A S Rizzo C Nunes B Pitigoi D Camara A L Mosnier A others 2011 Estimates of pandemic influenza vaccine effectiveness in Europe 2009 2010 results of In
48. ons force of infection used in D4 3 Prototype Software Documentation TELL ME project GA 278723 the epidemic model The arrows in this logic diagram therefore indicate the main interactions for which rules have been constructed Figure 1 Broad simulation logic gindviandsWavsucaareendssetusnacebikesuciereseeseauncenes Severity case fatality Regions Epidemic Demographic Personal health factors Ty a Base attitude Problem events Population density Attitude i _ Threat Susceptibility Riis Susceptible J population roon o S adopting Protective o Incidence Behaviour zi S 4 Baen TS Force of infection Migration Subjective Norms Communication plan snnnnnnnnnonnnnnnnnnnsnnnnnnnnnnnnnnnnnnnnonnnnnnnnnnnnn individuals Protective behaviour Efficacy The outcome of these model rules depends on specific characteristics of the simulated individual and their situation For example individuals will also have access to different information and communication based on their access to media channels Those information differences contribute to different behaviour 1 3 How we envisage you will use the simulation The TELL ME model was built as a prototype decision tool But it can also be used for other purposes At least four important uses have been identified during stakeholder discussions Exploring and comparing scenarios perhaps the most intuitive use of the simulation is to ru
49. or each repetition of each parameter combination that is each experiment or run In this case step is the tick at which the simulation ended If you set up the experiment to record results at each time step there will be multiple lines for each experiment and step indicates the tick for which the data applies The following columns contain the output These are the results of the simulation to be analysed You may wish to perform some tidying of the csv file but the file can be used without such processing The data can be directly analysed with spreadsheet software or imported to any data analysis software e g SPSS Stata R 4 1 3 Running headless NetLogo also allows you to run BehaviorSpace experiments headless i e from the command line on your operating system This means you won t use the graphical user interface and thus won t see the map view at all This is useful for automating runs on a single machine or a cluster of machines Whilst familiarity with using the command line is of course helpful it is possible to use this feature without prior experience Guidance can be found at https ccl northwestern edu netlogo docs by going to the BehaviorSpace page and scrolling to the section on running headless 4 2 Changing the simulation NetLogo is an accessible programming language While the TELL ME model is relatively complicated there are some changes that could be made by inexperienced coders This
50. ou want to undo but can t simply close and reopen the simulation without saving Even if you have saved some unwanted changes you can return the simulation to default parameter values and then save it again This section provides some guided play with a series of scenarios that gradually build complexity to assist you to learn to manipulate the simulation and understand outputs These are unrealistic scenarios designed to demonstrate how the simulation works and allow users to build a solid foundation of familiarity before they use it fully We shall now walk through the six scenarios summarise in Table 1 These instructions will begin with plenty of text to explain things but will become more succinct as you go through You may need to refer to Section 2 4 for a description of each new parameter as it is used If possible keep a note of the various outputs to compare results of the different scenarios if you are very unsure how to interpret the outputs of the simulation refer to Section 2 5 The scenarios only take a few minutes each but should be run in one session as they are incremental That is changes to parameters in each scenario assume that the changes already made in previous scenarios are still in effect 13 D4 3 Prototype Software Documentation TELL ME project GA 278723 Table 1 Learning scenarios Scenario Description 1 Epidemic Curve This scenario gives a random baseline epidemic with no communic
51. oward nonvaccination protection starts slightly higher than that towards vaccination There is no change in the attitude figures until the epidemic is officially declared at tick 54 This is because none of the messages were active before the start of the epidemic Once the epidemic has begun we see an increase in attitude as the messages have an effect but then a decay as time passes after the messages There is no clear difference in attitudes between the two scenarios suggesting the regular mass media message common to both scenarios is the most important component of the more complex communications plan with respect to effect on attitudes 2 5 2 Behaviour plot The behaviour plot shows the proportion of the population adopting vaccination and nonvaccination protective behaviours through time Figure 11 Behaviour Plot Complete Comms Plan Figure 10 Behaviour Plot One Msg Comms Plan Behaviour Behaviour E Protect E P Protect W vaccinate P Protect E P Protect E Vaccinate Adopting o m 2 Q D lt D4 3 Prototype Software Documentation TELL ME project GA 278723 We can see in Figure 10 and Figure 11 that the proportion of those exposed and infected adopting nonvaccination protective behaviours red line stays at zero until the epidemic begins then rises in both scenarios though higher in the more complete communications plan scenario Then after dropping a little rises to one and
52. parameters Travel parameters are used to estimate the proportion of new cases created by a patch that should be created in neighbouring and distant patches The parameter values of 0 25 travel rate and 0 6 travel proportion short are based on United Kingdom distance to work data National Centre for Social Research 2012 As well as differences in travel rates between countries these values are affected the size of the country being modelled because that influences how much area each patch represents If the travel rate is high and the travel proportion short is low an epidemic with low RO can easily die out before getting started because new infections are created in many new places and there is no central pool of infected people to provide a critical mass Therefore it is important to assess the realism of the spatial soread of the epidemic when adjusting these parameters rather than simply relying on data Three parameters are used to define the proportion of the population in specific categories Those with access to social media Prop social media is set at 0 7 a moderate value for European countries Eurostat 2014 We Are Social 2014 The number of healthcare workers per thousand population popn hcw is set at 10 International Labour Organisation 2014 Trading Economics 2015 The proportion with anti vaccination views was set at 0 1 ME 2012 All of these parameters are on the main interface and therefore able to be altered per
53. pidemic actually when National prevalence prevalence reaches half the level at which an epidemic is declared After and is intended to represent preparatory communications Day Regular Level gives the number of messages that occur before the epidemic ii at the start of the epidemic iii on a particular day after the start of the epidemic Day Level gives the day iv when the epidemic reaches a certain national prevalence Day Level sets the proportion prevalence v after the peak of an epidemic Day Level sets the relative prevalence compared to the epidemic peak vi repeated at regular intervals once the epidemic has started Day Level gives the number of days between repeated messages 19 Parameter Day Level Target Group Media Channel Content Behaviour Type Values User number input All High risk Health care workers Infected Anti vaccination Mass media Social media Health media Health profession Epidemic status Promote benefits Emphasise responsibility Recommend adoption Vaccination Both Other protective D4 3 Prototype Software Documentation TELL ME project GA 278723 Description Gives the number of days or proportion level to be used by the trigger see above This parameter identifies to whom messages are targeted and thus who they reach They can be targeted at i all of the population ii only high risk groups e g old young frail iii h
54. plan scenarios There were insufficient data available to set the parameters concerning the effectiveness of communication Lin et al 2014 Odone et al 2014 These parameters control model rules such as the amount that a person s attitude changes in response when they receive a relevant message Some of these parameters are set so that they have no effect for example trust is set to 1 but are available for users to adjust to enable experimentation or customisation following new research Others are arbitrary but are coherent with respect to other parameters For example each media channel has a parameter that controls how much a message recipient s attitude changes in the Effect of communication section of the Advanced Parameters screen and the values were chosen so that the stronger effect occurs with the more trusted information sources Maurer et al 2010 The values for accessing various media channels are entirely arbitrary However the Go to doctor during campaign parameter is informed by the literature Rubin et al 2010 The predefined communications plans scenarios high medium and low are based on discussions with project partners and other stakeholders of a draft developed from literature Rubin et al 2010 Lam and McGeer 2011 Department of Health UK 2012 Crosier et al 2014 27 D4 3 Prototype Software Documentation TELL ME project GA 278723 3 3 Population structure
55. re than 5 of the population have been or are infected during the epidemic Yy Incidence the total number of people from the population who have been newly infected during the current time step tick Max I the highest incidence during the current simulation When during which time step the highest incidence occurred VV Y Max P the highest prevalence during the current simulation and gt When during which time step the highest prevalence occurred Finally at the top right of the screen we see When Declared which displays when the epidemic in the current simulation was declared the time step at which the prevalence reaches a predefined level 2 2 6 Saving the simulation input values Any changes you make to the inputs on the main interface for example the country selected can be saved by simply saving the simulation When it is next opened it will have those different values selected The buttons along the bottom allow some of these changes to be reversed by restoring the default values that are included in the original model file 2 3 Learning to use the simulation The easiest way to learn how to use the simulation is to play with it Run it make a small number of changes and run it again to see what happens If you want to create a separate copy of the model to play with open the model and then save it under a different name File gt Save As and then open the file with the new name If you make any changes y
56. rotect to 0 Run the simulation that is press the setup button and then the go button While the model is running observe the world view map You should see the white dots becoming purple particularly near the most active epidemic areas The purple dot indicates that at least half of the agents on the patch have adopted protective behaviour Observe the attitude plot it should again be horizontal lines as there is no communication to change people s attitudes On the other hand the behaviour plot should show some patterns Most likely the blue line increases sharply at the beginning of the simulation settles for a while then increases during the active epidemic and decreases again as the epidemic dies away This is people reacting to the epidemic by adopting nonvaccination protective behaviour The black line probably increases much later and then Stays at the higher level This maintenance is because people who seek vaccination are vaccinated and 15 D4 3 Prototype Software Documentation TELL ME project GA 278723 cannot later become unvaccinated The red line is extremely volatile as it is based on the behaviour of the small number of agents who are currently infected Observe the epidemic plot and the results in the various epidemic monitors These should be identical to the first scenario epidemic results Why Because despite people adopting protective behaviour the protective behaviour doesn t provid
57. rs and scenarios in the simulation essentially the inputs we should consider how to use the outputs or results of the simulation This section will present the three plots map view and monitors that make up the outputs of the simulation in the interface and compare two differing scenarios for each to demonstrate what the outputs show The two scenarios used are exactly the same except that one has only a very small communications plan of one regular mass media message whereas the other has a more complete communications plan made up of a range of different messages It is also possible to compare many different scenarios repeat scenarios and take averages and use more quantitative analysis techniques NetLogo provides a tool to do this called BehaviorSpace See section 4 1 for details 23 D4 3 Prototype Software Documentation TELL ME project GA 278723 2 5 1 Attitude plot The attitude plot shows over time the average black and mid blue and plus and minus light blue and grey one standard deviation values for individuals attitudes towards vaccination and nonvaccination protective behaviours Figure 8 Attitude Plot with Complete Comms Plan Figure 9 Attitude Plot with One Msg Comms Plan Attitude Attitude C SD NV C SD NV E Ave NV f E Ave NV C SD NV O SD NV E sp V E sp V MB Ave v HB Ave v E sD V i E sD V We can see in Figure 8 and Figure 9 that the average attitude t
58. s are increased so that behaviour is less likely to be adopted The following two scenarios will then show how communication can lead to reaching those higher thresholds and greater adoption Setup instructions e Set protectV threshold to 0 35 e Set protectNV threshold to 0 35 Run the simulation remember to press setup first The results should have lower or even no behaviour adoption compared to scenario 3 and therefore less control of the epidemic 2 3 5 Scenario 5 Basic communication This scenario introduces communication Two messages are included one that affects attitude and hence behaviour and another that affects behaviour more directly This latter message adds a bonus to one of the inputs to the behaviour score increasing the score and making it more likely to exceed the threshold 16 D4 3 Prototype Software Documentation TELL ME project GA 278723 Setup instructions gt Switch to the Communications Plan screen gt For message 1 that is the top line Set Trigger m1 to Regular Day Level m1 to 10 Target group m to All Media channel m1 to Mass media Content m1 to Promote Benefits and Behaviour type m1 to Other Protective gt For message 2 the second line Set Trigger m2 to National Prevalence Day Level m to 0 01 Target group m2 to All Media channel m2 to Social media Content m2 to Emphasise Re
59. s occur but yesterday s new cases count more than the previous day s and so on If the weighted average of these three components exceeds some threshold the person adopts the relevant behaviour The other layer is a spatial epidemic simulation Mathematical equations are used to estimate the number of newly infected people in a region in the next time step based on the existing number of infected and susceptible people The layers interact with outputs from each layer influencing behaviour in the other layer One of the three elements in an individual s behaviour decision is based on the number of recent nearby new infections In the other direction the adoption of protective behaviour by the individuals in a region affects the spread of the epidemic in that region These model rules are fully described in report D4 2 While some details were changed as the model was constructed the general approach has not been altered Updated design details are not available at January 2015 the end of the TELL ME project but will be available in the academic literature The broad logic of the simulation is detailed in Figure 1 The key entities within the simulation are communication plans individuals and regions The figure describes major properties of these entities which are grouped by entity The arrows identify the pattern of influences between properties of entities For example the protective behaviour adopted by individuals affects the regi
60. s scenario there is no communication so attitude should not change This is visible in the attitude plot with all lines horizontal because nothing changes over time Furthermore the thresholds for behaviour adoption have been set to their maximum and agents can t adopt protective behaviour This is visible in the behaviour plot with everything at O for the entire run However there should be a curve in the epidemic plot as the epidemic spreads and infects more people but then dies out when most people have become immune Record the details of the epidemic for comparison with later scenarios Impact Max I and its associated When and Max P and its associated When 2 3 2 Scenario 2 Ineffective The purpose of the second scenario is to show adoption of behaviour returning the threshold parameters to their default values The behaviour is ineffective so it should not influence the epidemic We also take advantage of the capacity to repeat a simulation by using the same random seed or sequence of random numbers with different parameters Setup instructions gt Uncheck the randomise checkbox on the Advanced parameters screen This makes the simulation use whichever random seed is currently in place which was randomly generated during the first scenario Set protectV threshold to 0 3 Set protectNV threshold to 0 25 Set efficacy vaccine to O VV VV Set efficacy p
61. section is intended to gt introduce the reader to NetLogo code and identify some available support resources and gt provide an example of a simple change adding another country to the simulation options 4 2 1 Resources for NetLogo coding Simulations built in NetLogo are mostly built by writing programming code in the Code tab of the modelling environment The interface is also used to create some inputs and outputs such as sliders and plots NetLogo code is very naturalistic allowing for beginners to get to grips with the code quickly The NetLogo website https ccl northwestern edu netlogo has many resources that will be invaluable gt The User Manual this contains a huge amount of information about all aspects of NetLogo including guides and a dictionary of NetLogo commands gt Resources page this is an excellent place to start looking for discussion boards model libraries and other resources relating to NetLogo gt Tutorials these are located in the User Manual pages There are several books about agent based modelling that contain NetLogo examples and gradually develop models Vidal 2009 Teahan 2010 Railsback and Grimm 2012 Other learning resources include 31 D4 3 Prototype Software Documentation TELL ME project GA 278723 video tutorials on Youtube search for NetLogo and a free online course offered by the Santa Fe Institute through their Complexity Explorer site 4 2 2 Adding
62. sponsibility and Behaviour type m2 to Vaccination gt Switch to the Advanced parameters screen and increase the Bonus for vaccination to 0 5 its maximum Run the simulation can you see the changes in attitude There should be a bump in the attitude plot blue line every 10 time steps There should also be an increase in the proportion adopting vaccination and that adoption starts when prevalence reached 1 of the population What else has happened Overall behaviour levels should be higher than the levels that occurred in scenario 3 because the communications plan has partially overcome the high thresholds for behaviour change The message 1 parameters mean that a message is broadcast with mass media every 10 days after the epidemic is declared That message is intended for everyone and promotes the benefits that is increase attitude toward of nonvaccination behaviour The message 2 parameters mean that a message is broadcast once using mass media when the proportion of the population infected reaches 1 That message is intended for everyone and emphasises a person s responsibility that is applies a bonus to the calculation of the behaviour score for becoming vaccinated 2 3 6 Scenario 6 Communication plan The final scenario has multiple communication messages Also as was done for scenario 5 the effect of communication is exaggerated by altering some of the advanced parameters to make the results more vis
63. stics but provide further depth and context to the simulation if desired These situations were identified as important contextual elements by stakeholders during the development of the simulation model Five different policy scenarios are available in the simulation each with the own set of user controllable parameters Isolation this set of parameters relates to the inclusion of the possibility of compulsory or voluntary self isolation of individuals who are infected There is a checkbox to turn this context on or off a slider to show the proportion reduction in travel resulting from isolation a slider to show the proportion reduction in contacts in an individual s home region a number input to determine at what national prevalence percentage the isolation context becomes active and a slider to determine how long the isolation policy operates after it has become active this is not the period that a person remains isolated Restrictions on access to vaccination these parameters allow for a delay in access to a vaccine e g it takes time to be developed or restrictions in eligibility for an available vaccine There is a checkbox to turn this context on or off a selector box to choose which type of individuals the restriction applies to and a slider to pick how long the delay occurs Misperception of risk these parameters allow for an error in individuals perception of the risk the epidemic poses There is a checkbox to turn this conte
64. t the protectNV threshold slider to 1 When you opened the simulation all the settings in the Communications Plan Policy Context and Advanced Parameters screens were set to their default values You can gt gt Confirm that that are no active messages Click on the Communications Plan tab at the top of the window to change screens and check that Trigger left dropdown box is set to NONE for all lines Confirm that there are no active policy scenarios Click on the Policy Context tab and notice that all checkboxes at the left of the screen are empty 14 D4 3 Prototype Software Documentation TELL ME project GA 278723 Run the simulation gt Click the setup button This transfers all the settings on the sliders to the internal model variables and creates a small number of infections to start the epidemic You will see the correct country map displayed with a small light red patch to show where the epidemic is starting gt Click the go button gt Initially let the simulation run at normal speed but move the slider to the fastest speed so that the run does not take too long to complete gt The simulation will take a few minutes to run and may take a long time to really get going or even die out if you have too low a value for RO gt The simulation will stop when the epidemic has finished or when you click go again Observe the simulation For thi
65. ulation is relatively simple but does require several steps Briefly these are gt Download and install the NetLogo software gt Download and install the eXtraWidgets NetLogo extension and gt Download and save the TELL ME model file and supporting GIS data You must first download and install NetLogo on your machine NetLogo is freely available under a GPL GNU General Public License license at https ccl northwestern edu netlogo Go to this website and follow the on screen instructions to download and install the most recent version 5 1 0 or later for your operating system Windows OS X or Linux The TELL ME simulation takes advantage of the eXtraWidgets NetLogo extension developed during the project Before you open the simulation in NetLogo you will need to download this and save the files in the appropriate folder To do so go to https github com CRESS Surrey the link is also available from the TELL ME and CRESS websites and follow the link to the eXtraWidgets page Once you have downloaded the extension place the unzipped folder and its contents in the extensions folder inside the NetLogo folder the location of the NetLogo folder will depend on where you saved it during installation see Figure 2 More detailed instructions on how to install the extension are available from the github site Figure 2 Saving the eXtraWidgets Extension eee NetLogo 5 1 0 eee extensions E EE xw gt 33 Bo mo Sov By
66. xt on or off a slider to determine the perceived risk 21 D4 3 Prototype Software Documentation TELL ME project GA 278723 of infection as opposed to the real observed risk a slider to determine increased perception of the severity of the disease and a slider to determine the duration of the misperception it begins when the epidemic is declared Media frenzy these parameters relate to the commonly observed media frenzy that occurs early in a possible epidemic and the resultant increase in protective behaviour despite low real risk There is a checkbox to turn this context on or off and a slider to determine what proportion of the population adopts protective behaviour based solely on the media frenzy instead of calculating their behaviour based on the usual combination of attitude proportion nearby who have adopted protective behaviour and proximity to the epidemic front Loss of trust in health communication this set of parameters allows for the possibility that individuals have lost trust in the messages they receive as part of the communications plan There is a checkbox to turn this context on or off a slider to determine the average trust in messages at the start of the epidemic a slider to determine the trust level once recovered and a slider giving the time for which it takes trust to recover While the loss of trust exists messages will be less effective in changing attitudes or behaviour 2 4 8 Advanced param
67. you want displayed are included in the data Test the new country by selecting it and then pressing the setup button The map should appear If this is successful save the NetLogo file 33 D4 3 Prototype Software Documentation TELL ME project GA 278723 BIBLIOGRAPHY Ajelli M Goncalves B Balcan D Colizza V Hu H Ramasco J J Merler S Vespignani A 2010 Comparing large scale computational approaches to epidemic modeling Agent based versus structured metapopulation models BMC Infectious Diseases 10 190 Bauch C T Lloyd Smith J O Coffee M P Galvani A P 2005 Dynamically Modeling SARS and Other Newly Emerging Respiratory Illnesses Past Present and Future Epidemiology 16 791 801 Center for International Earth Science Information Network CIESIN Columbia University Centro Internacional de Agricultura Tropical CIAT 2013 Gridded Population of the World Version 3 GPWv3 Population Density Grid Future Estimates 2015 Centers for Disease Control Prevention 2012 Epidemiology and Prevention of Vaccine Preventable Diseases Centers for Disease Control and Prevention Cowling B J Ng D M lp D K Liao Q Lam W W Wu J T Lau J T Griffiths S M Fielding R 2010 Community psychological and behavioral responses through the first wave of the 2009 influenza A H1N1 pandemic in Hong Kong Journal of Infectious Diseases 202 867 876 Crosier A McVey D French J

Download Pdf Manuals

Related Search

Related Contents

Philips DVP3166K DivX DVD player with USB  Kyocera Finecam L3v User's Manual  RoverVu Module (Version 1.1)  第53回定時株主総会招集ご通知  Newstar KVM extender, UTP, USB  Dell SupportAssist versión 1.1 para Microsoft System Center  ICM-8D Fuel Injector Cleaner User Manual  Samsung E1130 Manuel de l'utilisateur  H7ー/TH7ーシリ一ズ 湿度/温・湿度検出器 〟取扱説明書  D-Power 1&05 取扱説明書  

Copyright © All rights reserved. Failed to retrieve file