Interpretable Deep Learning from Electronic Health Records under Learning Constraints

Patient and population data from various sources within Electronic Health Records (EHR) are becoming increasingly important for data-driven decision support and diagnostic assistance. This includes medical images. The development of interpretable AI is a crucial component for building trust in clinical decision support and diagnostic aid. This project specifically explores new interpretable methodologies within deep learning. In 2023, we have published several innovative methods. We have developed a method called "Relax," which provides users of an image analysis system with a visualization of pixels in input images that are most "responsible" for the decision suggested by the deep neural network. We have further developed this new method for the article "A clinically motivated self-supervised approach for content-based image retrieval of CT liver images," where the purpose is to interpret the importance of pixels in CT images in the context of liver segmentation for colorectal cancer. This work has been done in collaboration with clinicians and validated by clinicians. We have also researched new methods for so-called unsupervised classification, where there is limited training data, by better leveraging different image modalities in a synergistic way. Elements of these new ways of performing deep learning have been published in the journal Medical Image Analysis for "multi-class medical image volume segmentation."

Patient and population specific data from heterogeneous Electronic Health Records (EHR) are becoming ubiquitous sources for data-driven decision and diagnosis support systems. Deep learning artificial intelligence technologies are emerging as the state-of-the-art for EHR analysis due to their ability to learn complex representations from raw clinical data to obtain strong predictive power combined with an inherent ability to accept multiple data types as input for heterogeneous data fusion. However, key problems and constraints for deep learning systems for health are their lack of interpretability, their inability to exploit vast amounts of unannotated patient data, and their hitherto inability to exploit contextual information to perform well in the low volume data regime, e.g. due to stratification. As a key solution, the DEEPehr project will develop interpretable deep learning predictive systems for a range of EHR input sources, focusing particularly on prediction and prevention of postoperative adverse events. Adverse events, such as infections, are potentially lethal, causing huge suffering for patients and huge costs for healthcare. DEEPehr will develop novel unsupervised and weakly supervised deep learning methodology to exploit the wealth of unannotated patient data for better quality of care, and will leverage the unique hierarchical nature of EHRs for utilizing contextual and prior information to extract new clinical knowledge from low data volumes. Project results and outcomes will impact DEEPehr's clinical stakeholders, and the potential to impact data-driven health and science beyond is great given the generic methodology development core of the project. DEEPehr is high risk because of the profound challenges and interdisciplinary nature of the endeavor, yet feasible due to the high quality of the team, the extensive mobility, and the top international collaborators, creating the synergy effects needed to reach the ambitious project objectives.