class: center, middle, inverse, title-slide # Modeling Malaria and COVID-19: Models, Applications and Challenges ## (Slides available at: bit.ly/bentoh-nu) ### Ben Toh ### 2020/12/16 --- class: inverse, center, middle # Self Introduction --- <img src="../assets/img/hidden_heavens.png" width="100%" style="display: block; margin: auto;" /> --- <img src="../assets/img/coral_restore.png" width="100%" style="display: block; margin: auto;" /> --- # Overview ### Part I Improving National Level Spatial Mapping of Malaria through Alternative Spatial and Spatio-temporal Models <img src="../assets/img/malaria-risk-bf-intro.png" width="90%" style="display: block; margin: auto;" /> --- ### Part II Interactive visualizer and simulator for decision support Guiding placement of health facilities using malaria criteria and interactive tools .pull-left[ <img src="../assets/img/covid_dashboard.PNG" width="1457" style="display: block; margin: auto;" /> ] .pull-right[ <img src="../assets/img/covidsim.png" width="1321" style="display: block; margin: auto;" /> ] --- ### Part III Agent-based Model of COVID-19 Transmission in Florida <img src="../assets/img/flcovid-RCHD.png" width="78%" style="display: block; margin: auto;" /> --- class: inverse, center, middle # Part I Improving National Level Spatial Mapping of Malaria through Alternative Spatial and Spatio-temporal Models --- # Background summary .pull-right1[ - Mapping of malaria prevalence based on national malaria survey data - Choices choices choices of models - Common approach: SPDE-INLA - Alternative: Generalized Additive Model (GAM) - Alternative: Gradient Boosted Models/Trees (GBM) - More choices: To include the past dataset or not (Spatial vs Spatiotemporal settings) ] .pull-left1[ <img src="../assets/img/spde-gamgbm-length.png" width="50%" style="display: block; margin: auto;" /> ] --- # Objectives - To determine if GAM and state-of-the-art machine learning method (e.g. GBM), under both spatial and spatio-temporal setting, can be good alternatives to the more complicated SPDE method - To determine if inclusion of past data is beneficial in modeling the current spatial distribution of malaria prevalence at the national scale --- # Data - Demographic Health Survey data from five countries - Spatial covariates: free and publicly available remote sensing or GIS products .pull-left[ <img src="../assets/img/map-chp1-5cty.png" width="1365" style="display: block; margin: auto;" /> ] .pull-right[ <img src="../assets/img/chp1-covars-layer.png" width="1365" style="display: block; margin: auto;" /> ] --- # Model comparison - 5 countries × 4 models × 2 settings - Spatial setting <img src="../assets/img/chp1-schema-sp.png" width="2028" style="display: block; margin: auto;" /> - Spatiotemporal setting <img src="../assets/img/chp1-schema-st.png" width="2029" style="display: block; margin: auto;" /> --- # Results - GAM and SPDE 👍 <img src="../assets/img/BF_LLMAE.png" width="60%" style="display: block; margin: auto;" /> --- # Results - GAM and SPDE 🙌 <img src="../assets/img/UG_LLMAE.png" width="60%" style="display: block; margin: auto;" /> --- # Results - Very small difference among top models (look at the axis) <img src="../assets/img/NG_LLMAE.png" width="60%" style="display: block; margin: auto;" /> --- # Results - GBM 😲 <img src="../assets/img/ML_LLMAE.png" width="60%" style="display: block; margin: auto;" /> --- # Results - GBM 😲 GAM 😨 <img src="../assets/img/MW_LLMAE.png" width="60%" style="display: block; margin: auto;" /> --- # Spatial vs Spatiotemporal setting <img src="../assets/img/Diff-MAE-pres.png" width="60%" style="display: block; margin: auto;" /> --- # Discussions .pull-left[ - No single best model, performance varies from setting to setting and country to country - SPDE is consistent, but doesn't gain much from incorporating past data - Can deteriorate when past and present spatial dependency are very different ] .pull-right[ <img src="../assets/img/BF_pred_map.png" width="1575" style="display: block; margin: auto;" /> ] --- # Discussions .pull-left[ GAM is good, fast and simple-to-use alternative, especially with more data But ... ] .pull-right[ <img src="../assets/img/MW_pred_map.png" width="1575" style="display: block; margin: auto;" /> ] --- # Discussions <img src="../assets/img/five_country.png" width="4800" style="display: block; margin: auto;" /> - GAM: Dismal performance in irregularly shaped countries - High perimeter to root area ratio: Malawi 8.7, Uganda 5.7 --- # Discussions - GBM unpredictable but generally fits well with more data available - 🏠💬 Fit multiple model or at least benchmark with GAM --- class: inverse, center, middle # Part II Interactive visualizer and simulator for decision support --- # Interactive tools - Bridging the gap between modelers and stakeholders - Enhance understanding of the models by "using" them - Informed decisions - Increasingly easy to create: No longer requires in-depth knowledge in Javascript and CSS <img src="../assets/img/valle-conbio-grab.PNG" width="1888" style="display: block; margin: auto;" /> --- # Placement of health facilities - Early diagnosis and treatment of malaria reduce death and transmission - Many factors contribute to access to healthcare - Distance or travel time to health facilities is important predictor to malaria prevalence *(e.g. Schoeps et al. 2011, Kizito et al. 2012)* --- # Bunkpurugu-Yunyoo District, Ghana .pull-left1[ - 1200 km `\(^2\)` , 150K populations - 2 urban centers, 8 health facilities - Multiple years of malaria surveys in 2010 - 2014 - Important predictors *(Amratia et al. 2019; Millar et al. 2018)* - Distance to health facilities (HF) - Distance to urban centers ] .pull-right1[ <img src="../assets/img/byd-simplemap.png" width="2519" style="display: block; margin: auto;" /> ] --- # Objectives Determine the optimal locations for new health facilities based on district-wide malaria criteria: 1. Overall malaria prevalence of children under 5, 2. Overall malaria incidence of all ages, and 3. Average travel time to nearest health facilities --- # Methods - Three years of high transmission season data (2010 to 2012) - ~ 5000 children under five - 71 to 80 villages per year - GAM with 5 predictors - Travel time to HF, distance to urban center, elevation, NDVI, log population density - Use Genetic Algorithm to find optimal locations given number of health facilities and criteria --- # Results See interactive visualizer and simulator created. - http://bit.ly/ben-hf-app --- # Take home messages - Different optimization criteria can produce very different results. - Importance of using multiple optimization criteria in decision analysis. - Decision analysis and interactive application are important tools for communicating models. --- # Other Applications <!-- ## MSAT vs MDA --> <!-- ```{r echo=FALSE, out.width="100%", fig.align='center'} --> <!-- knitr::include_graphics("../assets/img/screen_shiny.PNG") --> <!-- ``` --> <!-- <small>Millar, J., Toh, K.B. & Valle, D. To screen or not to screen: an interactive framework for comparing costs of mass malaria treatment interventions. BMC Med 18, 149 (2020).</small> --> <!-- --- --> ### Predict Causes of Childhood Febrile Illness Symptoms, demographic and hematological variables as predictors. Bayesian Model Averaging approach. (http://bit.ly/ben-afi-app) <img src="../assets/img/afi_shiny.PNG" width="2321" style="display: block; margin: auto;" /> <!-- --- --> <!-- # Section remarks --> <!-- - Correlation `\(\ne\)` causation --> <!-- - Emphasis on "usability": e.g. small, fast, flexible --> <!-- - "Modeler-initiated" approach, does it work? --> --- class: inverse, center, middle # Part III Agent-based Model of COVID-19 Transmission in Florida --- # Hladish Lab <img src="../assets/img/hladishlab.png" width="80%" style="display: block; margin: auto;" /> --- # COVID-19 Agent-based Model for Florida <img src="../assets/img/abm_schema.png" width="1739" style="display: block; margin: auto;" /> --- # Use of ABM - Control and reopening strategy (MIDAS Multi-Model Outbreak Decision Support; submitted to *Science*) - Effects of vaccination (efficacy and coverage) - Spatial distribution and urban-rural divide of COVID-19 --- # Two Dynamics <img src="../assets/img/flcovid-RCHD.png" width="78%" style="display: block; margin: auto;" /> --- # Two Dynamics <img src="../assets/img/abc_schema2.png" width="1807" style="display: block; margin: auto;" /> --- # Reporting - Quantify relative improvement in case detection - Excess death, hospitalization-death dynamic varied over time <img src="../assets/img/covid_preprint.PNG" width="1376" style="display: block; margin: auto;" /> --- # Personal Protective Behaviour (PPB) - Important part of COVID-19 transmission dynamic - Using a simple 0-1 score to describe behavioural changes - Intertwined with mobility, policy intervention, social distancing, hand washing, masking etc - Affects disease transmissibility and inter-household interaction network in the model - Time-varying, need to parameterize using proxy such as “mobility index” --- # SafeGraph Mobility .pull-left1[ - Provides estimates of foot traffic to “point of interests” down to census block group level - Social distancing metrics, e.g. proportion of time at home 🏠 ] .pull-right1[ <img src="../assets/img/SG_TAH_FL.png" width="1889" style="display: block; margin: auto;" /> ] --- # Problem... <img src="../assets/img/case_vs_sgnah.png" width="70%" style="display: block; margin: auto;" /> - No clear change in behaviour during second wave - How did second wave go down without PPB change? - Herd immunity? --- # Search for other PPB proxy - Cuebiq contact index <img src="../assets/img/cuebiq_cci.png" width="1707" style="display: block; margin: auto;" /> --- # Search for other PPB proxy - Bars and restaurants “index” - Mask wearing behaviour: e.g. Imperial College YouGov Behaviour tracker - Composite index? --- # Work in progress... - Disentangling the interaction between mobility and age, and incorporating them in the model <img src="../assets/img/SG_TAH_Age.png" width="80%" style="display: block; margin: auto;" /> --- # References - Schoeps, Anja, et al. "The Effect of Distance to Health-Care Facilities on Childhood Mortality in Rural Burkina Faso", American Journal of Epidemiology 173, 5 (2011), pp. 492--498. - Kizito, James, et al. "Improving access to health care for malaria in Africa: a review of literature on what attracts patients", Malaria Journal 11, 1 (2012), pp. 55. - Weiss, D.J. et al. 2019. Mapping the global prevalence, incidence, and mortality of Plasmodium falciparum, 2000–17: a spatial and temporal modelling study. Lancet 394: 322–331. - Mandal, S. et al. Mathematical models of malaria - a review. Malar J 10, 202 (2011). - Millar, J. et al. 2018. Detecting local risk factors for residual malaria in northern Ghana using Bayesian model averaging. Malar. J. 17: 343. - Amratia, P. et al. 2019. Characterizing local-scale heterogeneity of malaria risk: A case study in Bunkpurugu-Yunyoo district in northern Ghana. Malar. J. 18: 1–14. --- class: inverse, center, middle # Thank you very much! Feedback, comments and questions? - Email: kokbent [at] ufl.edu - Slides: https://bit.ly/bentoh-nu - Website https://bentoh.my