Machine learning tools can model the effects of a meal on oral drug absorption

Written by Francesca Gavins

The Food Effect

Taking certain drugs with a meal can affect the safety and efficacy of the drugs, termed the food effect. Several tools exist in the lab to predict the effect of food on drugs, including simulated fluids and animal models. However, most are poorly predictive. Furthermore, drugs are investigated on a case-by-case basis and the data can be hard to interpret.

Machine learning has rapidly evolved with advanced technologies able to recognise patterns in data. Integrating machine learning into early food effect studies could help to streamline lengthy and costly studies.

Using machine learning

Here, we developed a bespoke machine learning pipeline using four different tools to predict the effects of food on oral drug absorption. We first identified a suitable dataset from the literature. We then tested several techniques in a systematic way. Random forest performed the best. High accuracies and sensitivities of 70%, 80% and 70% and specificities of 71%, 76% and 71% were achieved for classifying; no food effect vs food effect, negative food vs positive food effect and no food effect vs negative food effect vs positive food effect.

Machine learning has provided powerful tools able to predict a multifaceted process of food and drug absorption and the opportunity to expedite oral drug development studies.

Read our research article here.