Covid-net Biochem: An Explainability-driven Framework To Building Machine Learning Models For Predicting Survival And Kidney Injury Of Covid-19 Patients From Clinical And Biochemistry Data

This study introduces COVID-Net Biochem, a versatile and explainable framework for constructing machine learning models to predict COVID-19 patient survival and the likelihood of developing Acute Kidney Injury during hospitalization. The framework comprises two phases: the first phase is a “clinician-guided design” phase, which preprocesses the dataset using explainable AI and domain expert input. The second phase is a “transparent model selection” phase, which designs and trains a diverse collection of machine learning models based on the selected markers. The proposed approach advances machine learning model design by seamlessly integrating domain expertise with explainability tools, enabling model decisions to be based on key biomarkers.