Towards Precision Health: Harnessing Multi-Modal Explainable AI
| dc.contributor.author | Heinemann, Tim | |
| dc.date.accessioned | 2026-02-04T14:25:14Z | |
| dc.date.available | 2026-02-04T14:25:14Z | |
| dc.date.issued | 2025 | |
| dc.description.abstract | Precision health requires AI systems that are not only predictive but also transparent, interpretable, and biologically grounded. This talk outlines a multi-modal explainable AI framework designed to integrate diverse biomedical data types for diagnostics, personalization, and prevention. A central focus is placed on biologically informed neural networks (BINNs), which encode prior biological knowledge directly into model architectures to improve interpretability without sacrificing performance. The DeepBINN case study for acute kidney injury classification illustrates how pathway-level representations enable mechanistic insight and clinically meaningful explanations of model predictions. In parallel, phenomics-based approaches for personalized immune rejuvenation are discussed, highlighting the value of combining imaging, molecular, and functional data. Overall, the presentation demonstrates how multi-modal explainable AI can bridge the gap between black-box machine learning and actionable biomedical insight, paving the way for trustworthy AI adoption in precision medicine. | |
| dc.identifier.citation | AI+X Summit, 2 October 2025 | |
| dc.identifier.uri | https://hdl.handle.net/20.500.12839/1961 | |
| dc.identifier.url | https://www.plusx.ai/ | |
| dc.language.iso | en | |
| dc.rights | CC0 1.0 Universal | * |
| dc.rights.uri | http://creativecommons.org/publicdomain/zero/1.0/ | * |
| dc.title | Towards Precision Health: Harnessing Multi-Modal Explainable AI | |
| dc.type | Conference / Workshop | |
| dc.type.csemdivisions | BU-R | |
| dc.type.csemresearchareas | Tools for Life Sciences |