️ 67: Smarter Signals — How Multimodal AI Boosts Genetic Prediction of Heart Disease

️ Episode 67: Smarter Signals — How Multimodal AI Boosts Genetic Prediction of Heart Disease In this episode of Base por Base, we delve into a major advancement in the intersection of artificial intelligence and cardiovascular genomics. A study by Zhou et al. (2025), published in The American Journal of Human Genetics, introduces M-REGLE (Multimodal Representation Learning for Genetic Discovery on Low-dimensional Embeddings), a novel AI framework designed to enhance genome-wide association studies (GWAS) by integrating multiple physiological waveform modalities such as ECG and PPG.M-REGLE jointly analyzes these complementary health signals using variational autoencoders to generate low-dimensional, uncorrelated embeddings, which are then used to uncover new genetic associations. This multimodal approach allows for more effective representation of biological data than traditional unimodal models, significantly improving the power of GWAS and the accuracy of polygenic risk scores (PRS), especially for conditions like atrial fibrillation.Compared to unimodal learning, M-REGLE identified 19.3% more loci from 12-lead ECG data and 13.0% more loci from ECG+PPG data. It also achieved superior PRS performance across several datasets, including UK Biobank, Indiana Biobank, EPIC-Norfolk, and the British Women’s Heart and Health Study. Notably, the embeddings generated by M-REGLE remained unsupervised yet were predictive of cardiovascular diseases, suggesting that the model captures meaningful physiological and pathological patterns from raw data alone. Conclusion:M-REGLE exemplifies the transformative potential of combining multimodal physiological data with deep generative models for genetic discovery. By capturing both shared and complementary information across modalities, this AI-driven approach opens new doors to understanding the genetic architecture of cardiovascular diseases and improving clinical prediction tools using data already available from wearable devices. Reference:Zhou, Y., Khasentino, J., Yun, T., et al. (2025). Applying multimodal AI to physiological waveforms improves genetic prediction of cardiovascular traits. The American Journal of Human Genetics, 112, 1562–1579. https://doi.org/10.1016/j.ajhg.2025.05.015 License:This episode is based on an open-access article published under the Creative Commons Attribution 4.0 International (CC BY 4.0) license – https://creativecommons.org/licenses/by/4.0/

There are no comments yet.

Please log in to write the first comment.