Seminar: Zhendong Wang, Information Science and Engineering, KTH
Seminar
Date:Wednesday 7 May 2025
Time:13.00 – 14.00
Location:Campus Albano, lecture room 27, house 4, level 2
Counterfactual Explanations for Temporal Data in Healthcare
Abstract
Recent advancements in machine learning (ML) models for temporal data have demonstrated high predictive performance, yet these models often remain opaque. Counterfactual explanations can provide actionable insights by suggesting input modifications to achieve desired outcomes. These explanations are advantageous to apply in critical domains like healthcare to improve clinical decision support. This talk will highlight two studies:
1) Generating counterfactual explanations for temporal data prediction with counterfactual constraints, e.g., local temporal and domain-specific constraints. These constraints are integrated to emphasize the importance of temporal features and the relevance of application domains, such as healthcare;
2) The ongoing project of applying explainable ML and counterfactual tools to enhance both diagnostic and management capabilities for asthma patient management in Sweden.