Although artificial intelligence has already shown a promise in cardiovascular medicine, most existing tools analyze only one type of data – such as electrocardiographies or heart images – that limit their clinical utility. The appearance of multimodal AIs, which merges information from multiple sources, now allows algorithms to mimic the holistic reasoning of cardiologists and to provide more accurate knowledge of the patient.
The review, led by Western China Hospital at Sichuan University and the University of Copenhagen, examined more than 150 recent studies. The authors show that the combination of complementary ways – for example, echocardiography with computed tomography or cardiac magnetic coordination with genomic – significantly enhances diagnostic performance. A neuronal transformer network that merges chest radiographs with clinical variables at the same time identified 25 critical pathologies in patients with intensive care, achieving an average committee (AUC) 0.77. In another study, the incorporation of cardiac magnetic resonance imaging with all the genome has revealed new genetic sites that affect the functioning of the aortic valve, opening the doors to targeted prevention strategies.
In addition to diagnosis, multimodal AI can improve the choice of treatment. Mechanical learning models that incorporate imaging, laboratory results and drug history successfully predicted which heart patients would benefit from the treatment of heart reconstruction, distinguishing “over-reproduction” from non-respondents. Similar approaches have recognized patients who are unlikely to benefit from the repair of the mitral-venue by saving unnecessary procedures. The review also reports “biodegradable videos” derived from AO extracted from usual echocardiographies that independently provide for the evolution of aortic stenosis, allowing the opportunistic risk to rotate without additional tests.
Continuous home surveillance is another border. Algorithms that merge data from portable, smartphone applications and electronic health files can detect early degradation and deliver automated workouts, possibly reducing hospital readings. The authors estimate that the extensive adoption of today’s multimodal AI could reduce the costs of cardiovascular health care by 5% -10% within five years through improved performance and fewer complications.
Despite optimism, review warns that data quality, bias and algorithmic transparency remain significant obstacles. Models trained in oblique databases perform poorly in underpinning ethnic or socio -economic groups, while the nature of deep learning “black box” complicates clinical confidence. Researchers require standardized data collection, federal learning platforms and techniques to accelerate safe translation into ordinary care.
Source:
Magazine report:
Yang, X., et al. (2025). Using multimodal artificial intelligence to promote cardiovascular disease. Precision clinic. doi.org/10.1093/pcmedi/PBAF016
