Title:  Guiding Neural Network Decision Making with Medical Insights

Committee: 

Dr. AlRegib, Advisor    

Dr. D. Wang, Chair

Dr. Calhoun

Abstract: The objective of the proposed research is to integrate clinical context into the decision-making process of machine learning systems. This will guide the medical process but also it will increase translation from academic research to deployment in clinical practice. Numerous studies have demonstrated that deep learning is highly effective at analyzing medical imagery. In addition to imagery, studies have also shown that clinical records have been a vital data source for algorithms to perform diagnostic evaluations. These successes highlight the positive impact machine learning applications can have on clinical practice. Nevertheless, translation of machine learning algorithms from academia into clinical practice is seldom. This is largely because physicians cannot understand and justify the process through which the system arrived at its diagnostic decision. To alleviate these human barriers to the adoption of artificial intelligence in healthcare, in this dissertation, we study the interactions between machine learning models and the medical insights physicians use on a daily basis. Equipped with an understanding of such interactions, we plan to develop new methods to incorporate medical insights from healthcare professionals for guiding a model's decision-making process in both active and multi-modal learning frameworks. Then, we test this new set of medical machine intelligence systems on real data in real settings.