A combined AI approach to biomedical data analysis: Knowledge representation reasoning, machine learning and explainable AI

In this talk, I will explore if and how two traditionally distinct fields of AI, that is, ontology engineering and machine learning can be combined to improve performance outcomes. Using real world examples from epilepsy neurological disorder, the talk will demonstrate the use of biomedical ontologies in machine learning workflows to address the critical challenge of feature engineering in multi-modal non-numeric phenotype data. Specifically, we will discuss how biomedical ontologies can improve the performance of machine learning models and the runtime performance of machine learning algorithms. Further, the talk will also explore the role of explainable AI in the context of analyzing electronic health record (EHR) data to understand the role of hematologic biomarkers in patients with sepsis.

For Zoom information, please contact Robert Sanders (sandersrl@missouri.edu).