Webinar Wednesday: Knowledge Graphs for Explainable-AI (XAI): Explaining Machine Learning Models for Predicting Organ Transplant Survival
We present a novel Neural-Symbolic Explainable-AI (XAI) framework, integrating state-of-the art semantic knowledge graphs and machine learning (ML) methods, to make the output of ML models transparent so they can be safely used in clinical settings. Current XAI methods explain the output of ML models simply in terms of feature importance. Our XAI framework generates health knowledge-driven decision paths, comprising clinical concepts and evidence drawn from the literature, to provide a clinically-interpretable trace of how the ML model has processed the input data to generate its output/predictions. We applied our XAI framework for organ transplantation decision support, where we generate and visualize decision paths comprising organ transplant concepts to explain the ML-model’s predictions of kidney transplant survival and donor-recipient matches, thus making complex ML models useful for organ transplantation.
Speakers
-
Prof. Syed Sibte Raza Abidi
NICHE Research Group, Dalhousie University, and Professor of Computer Science and Medicine, Dalhousie University
Read More
Prof. Syed Sibte Raza Abidi, NICHE Research Group, Dalhousie University is Professor of Computer Science and Medicine at Dalhousie University, Halifax, Canada. He is the scientific lead of the NICHE (kNowledge Intensive Computing for Healthcare Enterprises) research group that conducts interdisciplinary research at the intersection of AI, health and digital health. He has over 25 years of research and innovation experience of AI-driven health data analytics and health knowledge management. He has developed over 40 AI-driven digital health solutions targeting clinical decision support, clinical guideline execution, digital therapeutics and personalized patient empowerment. He has published over 300 peer-reviewed research papers at top venues, secured over $18 million in research grants and supervised over 100 graduate students. He is the recipient of the Canadian Health Informatics Leadership Award, International Award for Innovation in Medical Informatics, Research Excellence Award and multiple best paper awards. He is the Board Member of International Association of Artificial Intelligence in Medicine and past board member of Digital Health Canada. Dr. Jaber Rad, NICHE Research Group, Dalhousie University
Read Less
-
Jaber Rad
Postdoctoral Fellow, Faculty of Computer Science, Dalhousie University
Read More
Jaber Rad is a Postdoctoral Fellow at the Faculty of Computer Science at Dalhousie University, where he recently completed his Ph.D. in the field of Explainable Artificial Intelligence (AI) and is a member of the NICHE research group. His research lies at the intersection of Semantic Web technologies and Machine Learning, with particular focus on generating explanations to interpret the predictions of complex ML models. He has published his research at international AI conferences. He also has expertise in visual analytics and is currently working on a Canadian Institutes of Health Research (CIHR) project to develop an analytical dashboard that monitors antibiotic consumption for promoting effective antimicrobial stewardship. Dr. Rad is passionate about advancing AI's reliability and interpretability in healthcare settings.
Read Less
