Explainable Machine Learning applied to Single-Nucleotide Polymorphisms for Systemic Lupus Erythematosus Prediction

Abstract

Systemic lupus erythematosus (SLE) is a type of autoimmune disease that affects multiple organ systems. The exact cause is unknown, but it is believed that predisposition to SLE is caused by multiple genetic factors. In this work we explored approaches to exploration and explanation of machine learning models for quantifying the risk of an individual to SLE using single nucleotide polymorphism (SNP) as features. Various model-agnostic explanation techniques were applied to further understand the factors that drive model predictions and allow comparison of the models. A web-based dashboard was developed to facilitate exploration and comparison of the models. The user can identify which features are important for predictions of each model, as well as to understand how a model comes up with a prediction for a given observation. The best performing model is the random forest model with AUC of 92.26% and AUCPR of 93.70g%.