Mapping tick dynamics and tick bite risk using data-driven approaches and volunteered observations

Presentation slides (Layman's talk, 15min) used during my PhD Defence, held on 27th of September 2019 at the University of Twente (the Netherlands). You can find the PhD thesis here: https://library.itc.utwente.nl/papers_2019/phd/garciamarti.pdf and a summary of the research in plain terms here: https://irenegarciamarti.com/post/rs00-explaining-my-phd-research/. Feel free to contact me at research [at] irenegarciamarti [dot] com if you have any questions.

Transcript

1. 1 Mapping tick dynamics and tick bite risk using data-driven approaches and volunteered observations Irene Garcia-Marti 27-09-2019
2. CONTEXT • Re-emergence of vector-borne diseases: • Global & socio-economic changes • VDB: 23 diseases (including Zika) • Changes + ticks • Longer tick season • Increase in tick populations • Northward + elevation expansion • New suitable habitats • Concurrent increase of tick-borne diseases Source: WHO
3. CONTEXT Source: RIVM Source: RIVM 25K – 27K LD cases per year ~2K develop persisting symptoms
4. CITIZEN SCIENCE TICK ACTIVITY • Monitoring tick activity • Project started and coordinated by WUR researchers • Monitoring 15-25 forested locations in the NL • Sampling: • Once a month, 2006-2016 • Using dragging blanket • Counting ticks every 25m • ~4,000 samples • One of the longest time-series on tick activity: unique!
5. CITIZEN SCIENCE TICK BITE REPORTS WUR + RIVM started Tekenradar in 2012 Currently 50,000+ samples
6. THE BIGGER PICTURE From “Lyme disease: The ecology of a complex system” R.Ostfeld (2012)
7. R = H x E RESEARCH OBJECTIVES Modelling tick bite risk and tick activity using citizen science data
8. DATA AND METHODS • General ML operations: • Classification • Regression • Pattern mining • ML methods: • Random Forest • Apriori • + • Statistics: • Poisson, neg. bin. • Zero-inflation, skewness
9. TICK ACTIVITY (H) Daily tick activity 15/4/2014 to 15/7/2014 Time-aware random forest 100+ covariates in 11 time windows
10. EXPLORING TICK BITE RISK What is the tick bites collection monitoring? Is there any information? EDA with Apriori to understand patterns of tick bites Tick bites are a realization of R!
11. HUMAN EXPOSURE (E) Pre-step before developing tick bite risk model Bridges the need of ICT data E clustered at the edges of forests and natural areas Locating E means we can characterize it R = H x E E = R ÷ H
12. 12 TICK BITE RISK (R) Modification of random forest to handle skewed and zero-inflated distributions Risk is higher along the coast Combination of H and 19 new covariates of E
13. CONTRIBUTIONS • Illustrating intrinsic relationship between R, H, E • Atmospheric water levels drive tick dynamics • Tick bite risk is strongly influenced by human exposure • Principles can be applicable to other tick-borne diseases Scientific • Modify method to handle time-aware observations • Modify method to handle skewed and zero-inflated distributions • Devised simple method to estimate human exposure to ticks without ICT • Guidelines to explore patterns of R Methodological • Monitoring elusive phenomena at unprecedented levels of detail • Observations can have sufficient quality to support scientific research • Sharing creative approaches at validating citizen science observations Citizen science
14. FUTURE WORK Improving prediction of tick dynamics Creating dynamic tick bite risk model Linking machine learning and statistics Creating apps for professionals and citizens
16. 16 THANKS research@irenegarciamarti.com