Mount Sinai Researchers Use Machine Learning to Assess CPAP's Impact on Cardiovascular Disease Risk in Sleep Apnea Patients

The study is notable for being the first to estimate whether continuous positive airway pressure (CPAP) therapy will increase or decrease an individual’s risk of cardiovascular issues. It highlights a shift towards precision medicine, offering tailored clinical approaches that may help mitigate cardiovascular risk in at-risk patients.

Obstructive sleep apnea affects approximately 25 million people in the United States and is linked to higher risks of cardiovascular diseases such as stroke and heart disease. CPAP is the primary treatment, delivering a continuous stream of pressurized air to maintain open airways during sleep. However, extensive prior research did not demonstrate a reduction in cardiovascular risks among CPAP users.

To create their predictive model, 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 utilized more than 100 predictors from sleep and health data to determine 23 critical baseline features for the analysis.

The findings indicated a significant variance in treatment responses among patients. The model identified a subgroup likely to benefit from CPAP, with those receiving the therapy experiencing a 100-fold reduction in cardiac risk compared to those receiving usual care. In contrast, another subgroup anticipated to be negatively impacted showed more than a 100-fold increase in adverse cardiovascular outcomes, like strokes and heart attacks, when using CPAP instead of standard care.

Co-corresponding author Neomi A. Shah, MD, emphasized that their research signifies a major advancement in personalized medicine, moving away from uniform treatment strategies for obstructive sleep apnea. The study underscored the promise of data-driven approaches to support clinicians in making informed treatment recommendations.

Co-primary author Oren Cohen, MD, acknowledged the predictive capabilities of machine learning but stressed the need for thorough validation before clinical application. Mayte Suarez-Farinas, PhD, highlighted the importance of advancing artificial intelligence in medicine beyond pattern recognition to causal reasoning, which is crucial for decision support tools in real-world scenarios.

Contributors to the SAVE trial included researchers from The George Institute for Global Health, University of New South Wales, University of Adelaide, and Flinders University in Australia. The study received support from various funding sources, including the Stony-Wold Herbert Fund, American Academy of Sleep Medicine Foundation, and the National Heart, Lung, and Blood Institute at NIH.