Researchers at Mount Sinai have developed a machine learning model capable of predicting the cardiovascular disease risk for patients suffering from obstructive sleep apnea (OSA). This innovative tool was recently detailed in a study published in Communications Medicine.
Read More
The study marks a pioneering effort to estimate how continuous positive airway pressure (CPAP), the primary treatment for OSA, may elevate or lower an individual's cardiovascular risk. Obstructive sleep apnea, affecting around 25 million Americans, poses significant cardiovascular risks, including stroke and heart disease. While CPAP is the most effective treatment to mitigate breathing interruptions during sleep, previous extensive studies have not conclusively demonstrated that it reduces cardiovascular risks.
Using a machine learning algorithm, the Mount Sinai team analyzed data from the Sleep Apnea Cardiovascular Endpoints (SAVE) trial, which involved over 2,600 participants across 89 sites in seven countries. They identified more than 100 health and sleep predictors to calculate 23 key baseline features, such as previous health issues and smoking habits.
The findings revealed significant variability in treatment responses across the patient cohort. The model highlighted a subgroup likely to benefit from CPAP, showing a 100-fold improvement in future cardiac risk when compared to standard care. In contrast, another subgroup was predicted to suffer harm from the therapy, facing over a 100-fold increase in adverse cardiovascular events, including recurrent strokes and heart attacks, with CPAP treatment compared to usual care.
Co-corresponding author Neomi A. Shah, MD, emphasized the importance of this research in advancing personalized medicine, allowing for tailored treatment approaches to meet individual patient needs. Co-primary author Oren Cohen, MD, noted the necessity of validating such predictive models for practical clinical application.
Mayte Suarez-Farinas, PhD, a co-corresponding author, underlined the shift needed in artificial intelligence in healthcare from merely recognizing patterns to understanding causal relationships, enhancing decision-making in real-world treatment scenarios.
Contributions to the study also came from The George Institute for Global Health in Sydney, Australia, among others. Funding was provided by various organizations, including the Stony-Wold Herbert Fund and the National Heart, Lung, and Blood Institute.
Mount Sinai Health System is a prominent academic medical network in New York, integrating various healthcare services and focusing on innovative patient care solutions.