Title: Integrating machine learning and symbolic AI to improve predictive models in healthcare
Speaker: Jorge Gaete
Abstract: The critical nature of medical tasks makes explainability an essential quality of any support system for this domain. Various techniques have been developed to provide explainable ML models for healthcare, nevertheless challenges still exist. Issues such as the integration of multiple sources of information or user interaction with the models are important areas of research to achieve more understandable models. In this talk we introduce our approach to tackling some of these issues by combining current explainable machine learning techniques and symbolic AI. As part of my second-year review, I will present a research pipeline and current work on its implementation and plans for future work. Special attention will be placed on the extraction of clusters as an approach to explain patterns of multimorbidity in patients and also the usage of logic programming to create a simple diabetes risk-predictor.