Artificial Intelligence and Medication Management

Abstract

Artificial intelligence in general, and machine learning (ML) techniques in particular, hold the promise of enormous benefits to medication management, both in the hospital or in the community setting. The potential of ML techniques to support the decision making of patients, clinicians, managers and/or policy-makers is massive. However, the learning will only be as good as the data, and the frame problem around medication is still to be addressed. While ML techniques offer a promising response to the various challenges in medication management, their implementation to help daily care faces many barriers. Data quality is key and must be improved, specifically at the point of capture (standardized data, shared data model, etc.), not only in electronic health records but also for all health-related information (e.g., home electronics). In addition, to fully exploit the potential of ML techniques in medication management, specific challenges need to be addressed to ensure that the tools based on these techniques are effective and can be deployed in daily care. This chapter will present key challenges that must be faced in the development and implementation of ML algorithms for medication management, specifically to estimate exposition to medication, as well as positive and negative outcomes associated with such exposition. Finding ways to describe and include the variability of exposure, and the uncertainty of reactions as part of the development of algorithms will be crucial to make sure the potential can unravel both at the individual and population level.