Approaching Complexity in OneHealth by Mathematical Modeling and Machine Learning

Abstract

The term OneHealth was coined approximately two decades ago and it is still conceptualized until this day. Most definitions have some common points, while exhibiting sometimes notable differences. Independently of these definitions OneHealth is an attempt to cover the complexity of several ecosystems coexistence such as human, animal, environmental, microbial, socio-economics, global health and governance ecosystems. Depending on these OneHealth definitions, the interactions between these ecosystems can in turn be defined with more or less detail. This paper will attempt to present the main models of interactions between ecosystems and the computational challenges and needs to achieve some consistent and quantitative results illustrated by some case studies such as the interaction between human, animal and microbial ecosystems leading to new infectious disease, the impact of the environmental ecosystem on human health through environmental risk factors etc. All these interactions need detailed and multiscale modeling, various computational strategies and massive input data to get insightful and relevant results. Furthermore, we argue that data (especially extracted from images, e.g. Earth observation images) should be extracted through automatic algorithms taking advantage of the newly developed machine and/or deep learning (ML/DL) strategies. Predictive models, as well should be developed based on these ML/DL techniques. It will quickly emerge that diversity and complexity of OneHealth issues need a series of methods as diverse and complex as the problems raised by this concept.