Machine Learning Model Interpretability for Precision Medicine

TL;DR

This paper demonstrates that complex machine learning models like random forests can be made interpretable for precision medicine, enabling better understanding of feature effects while maintaining high predictive accuracy.

Contribution

It applies model-agnostic explanation techniques to complex models in healthcare, showing interpretability without sacrificing accuracy.

Findings

Achieved 80% balanced accuracy in ICU mortality prediction.

Successfully interpreted feature effects at the individual patient level.

Proved that complex models can be made interpretable in medical settings.

Abstract

Interpretability of machine learning models is critical for data-driven precision medicine efforts. However, highly predictive models are generally complex and are difficult to interpret. Here using Model-Agnostic Explanations algorithm, we show that complex models such as random forest can be made interpretable. Using MIMIC-II dataset, we successfully predicted ICU mortality with 80% balanced accuracy and were also were able to interpret the relative effect of the features on prediction at individual level.