Modeling Campylobacter infection dynamics and estimating immune parameters from the MAL-ED study using Approximate Bayesian Computation

Abstract T15

Presenter: Nitya Singh (University of Florida)

Authors: Nitya Singh (1), Arno Swart (2), James Platts-Mills (3), Arie Havelaar (1)

• 1 Emerging Pathogens Institute, University of Florida, Gainesville, FL, USA;
• 2 Centre for Infectious Disease Control, National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands
• 3 University of Virginia, Charlottesville, VA, USA

Previous efforts of modeling Campylobacter infection dynamics describe the effect of boosting and waning of immunity in exposed populations. Those, however, have been developed using experimental challenge-rechallenge study data only. For estimating the immunity parameters we used the recently published MAL-ED study data on Campylobacter status, both serology and PCR, in diarrheic and non-diarrheic children from low-resource settings. The previously proposed infection dynamics model by Swart et. al. 2016 was extended and MAL-ED data was used for model fitting. We propose a dynamic age-dependent model of immunity using the ABC (Approximate Bayesian Computation) algorithm at its core for simulation and optimizing the parameter estimation. Data from Bangladesh, India, Nepal, Peru, and Pakistan, were used for this study, and immune parameters were estimated for Campylobacter jejuni/coli (CjCc) and Campylobacter species (Cspp), using data collected via PCR and ELISA based detection. For CjCc the estimated force of infection was in the range of 3-7 year-1, while for Cspp, it was found to be in a little higher range of 5-20 year-1. A similar trend was observed for the estimated duration of full protection being 0.036-0.082 years and 0.054-0.084 years for CjCc and Cspp respectively. Duration of partial protection was found to be higher for both CjCc and Cspp, in the range of 0.16-0.32 years and 0.07- 0.25 years. Incidences of Campylobacter-associated diarrhea (year-1) were found to be in the range of 1.0-2.3 and 1.7-3.7 for CjCc and Cspp respectively. These preliminary estimates are higher than MAL-ED reported incidence, while providing support to our hypothesis that standard estimates are biased due to their inability to exclude the protected people in the cohort. Further evaluation and cross-validation of the estimates are ongoing.

The proposed ABC model seems promising to handle person-level data for immunity model predictions. It can be utilized to predict the impact of changing the exposure frequency and dose on the prevalence of Campylobacter in children, to support hygiene interventions in low-resource settings. It may provide novel opportunities to estimate the incidence of pathogen-specific diarrheal illness in high exposure populations and yields insight into the mechanisms of waning and boosting. However, currently due to the unavailability of data confirming the immune status of the population cohort, the model fitting was quite complex and computationally expensive. So, the current findings also suggest the need for more serological testing to confirm the partial or full protection status of the selected population cohort and allow better fitting of the model for estimating the illness incidences in the population.

Keywords: Campylobacter, exposure, infection, illness, immunity models, ABC algorithm, Diarrheal incidence, young children.

About the presenter

Dr. Nitya Singh is an Assistant Scientist at the Department of Animal Sciences and Emerging Pathogen Institute, University of Florida. Before taking this position, she has received post-doctoral training at the University of Florida at Havelaars Lab. Dr. Singh holds a BTech degree in Biotechnology and an MTech degree in Information Technology with a concentration in Bioinformatics, and a Ph.D. in Bioinformatics from the Indian Institute of Information Technology Allahabad, Inida. She got trained in Infectious disease modeling with ICI3D (International Clinics on Infectious Disease Dynamics and Data) and received graduate training in the Public Health and Biomedical Informatics at UF College of Medicine. Her research focuses on genomic epidemiology, phylodynamics, meta-/genomics, and statistical data analysis directed towards tracking molecular links for possible outbreaks and illnesses. She is involved in infectious disease modeling, attribution of human disease to food and other pathways, quantitative microbial risk assessment-related research. She is also part of projects focusing on food safety and women empowerment in low- and middle-income countries. Her current research is funded by the Bill & Melinda Gates Foundation and International Livestock Research Institute(ILRI) and International Development Research Centre (IDRC).

Presenting in Speaking session 3 - Epidemiology and public health