AASM highlights opportunities, weaknesses of AI in sleep medicine

The Artificial Intelligence in Sleep Medicine Committee of the American Academy of Sleep Medicine (AASM) recently conducted a strategic analysis of AI’s advancements in sleep medicine and assessed how these technologies could improve care.

“AI is disrupting all areas of medicine, and the future of sleep medicine is poised at a transformational crossroad,” stated lead author Anuja Bandyopadhyay, PhD, chair of the Artificial Intelligence in Sleep Medicine Committee, in a press release. “This commentary outlines the powerful potential and challenges for sleep medicine physicians to be aware of as they begin leveraging AI to deliver precise, personalized patient care and enhance preventive health strategies on a larger scale while ensuring its ethical deployment.”

The authors emphasized that sleep medicine is well-positioned to take advantage of AI advancements due to the wealth of physiological data present in patients’ EHRs, much of which is obtained through sleep studies or wearable devices.

The authors posited that by using this data, AI could revolutionize sleep medicine across three domains: clinical applications, lifestyle management and population health.

The authors indicated that in the clinical setting, AI tools can play a pivotal role in data analysis and pattern recognition, which can support automation in diagnosis and clinical decision-making for chronic, sleep-related conditions. Further, AI deployment could improve efficiency in clinical workflows and bolster patient access, leading to reduced clinician burnout.

AI’s potential in lifestyle management stems from its integration into consumer sleep technologies, such as fitness trackers, smart rings and smartphone apps. Wearable sleep technologies can contribute to better sleep health by allowing patients to track and assess their sleeping behaviors. A recent AASM Sleep Prioritization Survey found that 68% of adults using sleep trackers reportedly changed their behaviors based on insights from the devices.

Despite the significant opportunities presented by AI-driven sleep trackers, the authors underscored that maintaining an ongoing dialogue between patients and providers about the limitations of these tools is critical.

AI’s ability to analyze vast amounts of data may also translate to improvements in population health.

“AI has the exciting potential to synthesize environmental, behavioral and physiological data, contributing to informed population-level interventions and bridging existing health care gaps,” Bandyopadhyay explained.

Alongside these opportunities, however, the commentary also raised important concerns around the deployment of AI in sleep medicine, including questions about bias, data privacy and security, accuracy and potential clinician over-reliance on the tools.

“While AI can significantly strengthen the evaluation and management of sleep disorders, it is intended to complement, not replace, the expertise of a sleep medicine professional,” noted Bandyopadhyay.

The authors also highlighted the importance of rigorous standardization and validation for healthcare AI tools, as these protocols help ensure the reliability and accuracy of the technologies upon deployment. This is particularly relevant for digital tools measuring sleep, as lack of standardization has led to discrepancies in the past.

“Our commentary provides not just a vision, but a roadmap for leveraging the technology to promote better sleep health outcomes,” Bandyopadhyay said. “It lays the foundation for future discussions on the ethical deployment of AI, the importance of clinician education, and the harmonization of this new technology with existing practices to optimize patient care.”

The commentary provides guidance for sleep medicine professionals who may be interested in AI as explorations of these technologies continue.

In March, researchers from Mount Sinai received a four-year, $3 million grant from the National Institutes of Health (NIH) to support the development of AI-enabled cardiovascular disease risk models for use in sleep apnea patients.

Obstructive sleep apnea puts patients at higher risk of cardiovascular disease, and while treatments like the use of continuous positive airway pressure (CPAP) machines can help lower this risk, additional risk reduction strategies are needed.

To that end, the Mount Sinai team will develop machine learning tools to predict which obstructive sleep apnea patients are at high risk for cardiovascular events like atherosclerosis progression and heart attack.