Development of digital decision-support tools to improve rehydration and antibiotic guideline adherence for diarrhoeal diseases

For over ten years, we have sought how best to develop decision-support tools for clinicians treating diarrhoeal disease. This is important because of a need to rapidly train large numbers of providers during cholera outbreaks, rapidly re-educate when guidelines change, address inappropriate antibiotic use, and allow for differences in epidemiology by season and place. We have built tools in both paper and digital formats, and evaluated their impact in clinical trials. Now, we are collaboratively building digital tools that depend on models developed through machine learning. In a large international collaboration, we have built and evaluated improved algorithms to assess dehydration for children and adults (project led by A. Levine at Brown University) and provide a probability that a patient has only a viral disease based on real-time weather, clinical and epidemiologic data (project led by D. Leung at Utah University). In addition to helping to improve cholera response, these tools represent a significant shift in how clinical decision-support might be in 10 years.

The impact of this research is to first improve rehydration and antibiotic guideline adherence. Secondly, the impact is to make possible dynamic decision-support that is responsive to where and when the patient is being treated.