Machine learning and AI for individualised therapy: Now and then

There was a time when we performed laborious pharmacokinetic hand calculations to tailor drug dosages and regimens – a practice that can still be useful for those brave enough to perform logarithmic calculator manipulations. These days, we are fortunate to have an increasing number of software tools to help us with the task, however many rely on population pharmacokinetic models typically built with relatively small data sets, and often necessarily representing subpopulations with niche requirements. Tomorrow, our work will no doubt be helped by artificial intelligence and machine learning approaches, and we have already started to see such approaches applied in the research arena.

Today, we are lucky to be joined by two experts from the field of pharmacokinetic and pharmacodynamic modelling, Prof Michael Neely and Prof Jean-Baptiste Woillard.

In this episode we'll hear about:

• the differences between traditional population pharmacokinetics and machine learning approaches in predicting drug exposure and optimizing dosing regimens

• some advantages of ML approaches, including in handling nonlinear patterns

• the challenge of interpretability of results and dose recommendations from machine learning methods compared to traditional pharmacokinetic methods

• which clinical contexts we will soon see machine learning derived models applied to clinical care

• the future of modelling and machine learning in pharmacology, including ethical considerations and the concept of augmented intelligence

Eight Drugs a Week is supported by the International Association of Therapeutic Drug Monitoring and Clinical Toxicology (IATDMCT). The personal views of the hosts and guests do not necessarily reflect the position of the Association.

To join IATDMCT, visit: https://iatdmct.org/join-us/

Subscribe to Eight Drugs a Week on your preferred podcast platform.

About our guests:

Jean-Baptiste is a Professor of Clinical Pharmacology in the Department of Pharmacology, Toxicology and Pharmacovigilance in the Limoges University Hospital, France where he leads the Pharmacometrics and Artificial Intelligence hospital functional unit. He is also Deputy Director of the Inserm Unit 1248 at the University of Limoges, France.

Michael Neely is a Professor of Pediatrics at the Keck School of Medicine of the University of Southern California, Chief of the Division of Infectious Diseases and Director of the Laboratory of Applied Pharmacokinetics and Bioinformatics, both at the Children's Hospital Los Angeles.