Explainable multimodal machine learning model for classifying pregnancy drug safety

Abstract

Motivation Teratogenic drugs can cause severe fetal malformation and therefore have critical impact on the health of the fetus, yet the teratogenic risks are unknown for most approved drugs. This article proposes an explainable machine learning model for classifying pregnancy drug safety based on multimodal data and suggests an orthogonal ensemble for modeling multimodal data. To train the proposed model, we created a set of labeled drugs by processing over 100 000 textual responses collected by a large teratology information service. Structured textual information is incorporated into the model by applying clustering analysis to textual features. Results We report an area under the receiver operating characteristic curve (AUC) of 0.891 using cross-validation and an AUC of 0.904 for cross-expert validation. Our findings suggest the safety of two drugs during pregnancy, Varenicline and Mebeverine, and suggest that Meloxicam, an NSAID, is of higher risk; according to existing data, the safety of these three drugs during pregnancy is unknown. We also present a web-based application that enables physicians to examine a specific drug and its risk factors. Availability and implementation The code and data is available from https://github.com/goolig/drug_safety_pregnancy_prediction.git. Supplementary information Supplementary data are available at Bioinformatics online.

embryoTox: Using Graph-Based Signatures to Predict the Teratogenicity of Small Molecules

“…Under stratified 10-fold cross-validation, our final model achieved Matthew's correlation coefficient (MCC) of 0.80, overall accuracy of 91%, sensitivity of 0.91, and specificity of 0.93, outperforming alternative proposed approaches 1,17 (Table 2).…”

“…Teratogenic drugs lead to malformation of the fetus and are strongly associated with lifelong physical and mental disabilities. 1 Significant effort, therefore, is invested in trying to avoid exposure to potentially teratogenic drugs during pregnancy. This was particularly brought to the public's attention in the 1960s, when the drug Thalidomide, approved as a sedative for treating nausea and vomiting during pregnancy, was found to have led to severe birth defects and deaths.…”

Predicting drug characteristics using biomedical text embedding

“…The modalities we evaluated are the different biomedical text embeddings, the drug’s molecular structure [ 29 ], and the DrugBank drug category. DrugBank’s drug category combines the drug’s ATC codes of different levels, and other manually curated drug information [ 30 ]. The DrugBank category is considered the most informative modality for drug safety classification, but it is not available during the early stages of a drug’s development.…”

“…The AUC and AUPR metrics, using 10-fold cross-validation were used for the comparison. The code published by [ 30 ] for feature engineering and model generation was used.…”

Developing an ensemble machine learning model for early prediction of sepsis-associated acute kidney injury

“…We constructed an ensemble supervised machine learning model based on the ‘stacking’ method, which refers to fitting multiple machine learning models on the same dataset and using secondary modeling to learn how to best combine their predictions ( Shtar et al., 2021 ). A single sub-model is called a first-level learner, while the combined model is called a second-level learner.…”