Machine Learning to Identify Population Specific Microbial Differences in Bacterial Vaginosis

While machine learning (ML) has shown great promise in medical diagnostics, a major challenge is that ML models do not always perform equally well among ethnic groups. Here, we investigated the ability of four ML algorithms to diagnose BV. We determined the fairness in prediction of asymptomatic and symptomatic BV using 16S rRNA sequencing data from Asian, Black, Hispanic, and White women.

Prediction was evaluated using balanced accuracy and average precision of four machine learning models (Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), Multi-layer Perceptron (MLP) classifiers) trained on the dataset.

General purpose ML model performances varied based on ethnicity. Feature selection, using t-test, revealed several bacterial species that varied in accurate prediction. These vary when investigating asymptomatic and symptomatic women. These models can be leveraged to elucidate bacterial species that contribute to adverse BV outcomes and inform novel therapeutic targets.