PTI, February 7, 2025: Artificial intelligence will be essential for diagnosing sleep disorders, such as sleep apnoea or insomnia, which are increasingly acquiring an individual character, neurologist Dr. Manvir Bhatia said.
Further, once diagnosed, are we able to say that a certain individual will be a good user of the treatment or not? That’s where the role of AI is really going to come up, said Dr. Bhatia, the vice-president of the Indian Society for Sleep Research.
She was addressing a workshop conducted at the All India Institute of Medical Sciences (AIIMS), New Delhi, ahead of the 11th Congress of the Asian Sleep Research Society that will be held here during February 8-9.
Obstructive sleep apnoea is a common disorder, estimated to affect about a billion people worldwide. Muscles in the throat narrow the airways, thereby affecting one’s breathing and blood oxygen levels.
A 2023 study by AIIMS estimated that 104 million Indians of working age suffer from obstructive sleep apnoea, of whom nearly 50 percent have it moderately or severely.
The first line of treatment usually includes PAP or ‘positive airway pressure,’ in which pressured air is delivered to the patient via a mask to keep the upper airway open, thereby preventing ‘apnoea’ or breaks in breathing.
“It’s okay that the technology will be used to say about the usage of PAP. But are we able to say that a certain individual will be a good user of the treatment or not, similar to how we are now able to predict one’s risk of having a stroke or a cardiac event? That’s where the role of AI is really going to come up—in predicting one’s PAP compliance,” said Dr. Bhatia, who runs the Neurology and Sleep Centre in New Delhi.
While sleep has been studied extensively over the years, the focus has largely been on AHI (apnoea-hypopnoea index) and sleep stages because the human eye could see only so much, according to the neurologist. AHI measures how often one’s breathing stops or slows down.
She added that as sleep technology becomes more compact and sophisticated, and the recorded information more granular and voluminous, artificial intelligence will be almost indispensable for analysing the data.
“There’s no way that the human mind can analyse all this data,” the neurologist said. “Sleep disorders are very heterogeneous in nature, be it sleep apnoea or insomnia—there must be something that we are missing. It’s never one treatment or diagnostic method that fits all, and that’s why we definitely need to incorporate AI into sleep medicine,” she said.
Dr. Bhatia also acknowledged the resistance of medical doctors in turning away from how sleep studies have been traditionally conducted toward newer technologies.
A traditional sleep study or polysomnography uses wires and sensors attached to a participant to measure physiological activity, including brain waves, breathing and heart rates, and oxygen levels during sleep. However, polysomnography is expensive and technology-, technique-, and technician-dependent, Dr. Bhatia said.
Further, the sensors on the body can interfere, and if you’re trying to assess someone’s sleep, are we really doing a good job? she added.
Home sleep testing devices, including patches, therefore, can be comfortable, convenient, and less tech-heavy, Dr. Bhatia said.
She also pointed out that consumer wearable technologies, including smartwatches, are increasingly highlighting the importance of sleep, even though the data may not be accurate.
“People are becoming very conscious. The devices are creating an awareness that sleep is important,” Dr. Bhatia said. However, she cautioned that these wearable devices are different from those used for medically diagnosing sleep disorders.
She also advised taking an overall, temporal view of the information provided by wearables, rather than diving deep into a night-by-night analysis.
Sleep apnoea is a highly prevalent condition, with serious comorbidities and consequences. “No way are we going to be able to diagnose so many people using traditional systems, even though they will continue to have a place,” Dr. Bhatia said in conclusion.
