Multi-Task Transformer Visualization to build Trust for Clinical Outcome Prediction

Dario Antweiler, Florian Gallusser, Georg Fuchs

Room: 104

2023-10-22T03:00:00ZGMT-0600Change your timezone on the schedule page
2023-10-22T03:00:00Z
Exemplar figure, described by caption below
Proposed visual analytics system that fosters trust into clinical transformer models, consisting of multiple interactive views: (1) Trust in dataset with feature distribution plots and coordinated hierarchical medical code visualization and co-occurance diagrams, (2) trust in model architecture & training with architecture diagram and training loss graph, (3) trust in validation with precision-recall/ROC curves & baseline benchmarks, and (4) trust in prediction with Shapley-values to display feature importance for individual predictions.
Fast forward
Full Video
Keywords

Outcome prediction, Explainable AI and ML in healthcare, Clinical workflows, Patient safety

Abstract

Clinical decision support systems based on machine learning are a rising application in healthcare. Early detection of deteriorating conditions provide the opportunity for medical intervention in hospital patients. Recent approaches increasingly rely on Large Language Models such as BERT, because patient data is often in the form of structured temporal data. These models are notoriously hard to interpret and therefore to trust, while precisely trust is an essential principle for technology in healthcare. We develop a visual analytics system to inspect, compare, and explain pre-trained transformer models for a given clinical outcome prediction task. The work is developed on the basis of a large hospital patient dataset and prediction tasks for acute kidney injury and heart failure. Discussion with healthcare professionals confirms that our system can lead to a faster decision process and improved modeling results.