AI can identify if you're at risk from more than 100 health conditions after just a single night's sleep, study says

Published on Jan 15, 2026 at 8:50 PM (UTC+4)
by Ben Thompson

Last updated on Jan 15, 2026 at 8:51 PM (UTC+4)
Edited by Emma Matthews

This new AI model can identify whether you’re at risk from more than 100 health conditions after just one night’s sleep.

SleepFM is a large language model (LLM) developed by researchers at Stanford University in California.

It can read a user’s brain activity, heart rate, respiratory signals, leg movements, and eye movements, all while they’re sleeping away.

From this data, it can predict the likelihood of whether or not a person will develop one of more than 100 health conditions.

EXPLORE SBX CARS – Supercar auctions starting soon powered by Supercar Blondie

How this AI model can predict your chances of developing a health condition – while you’re sleeping

In a study published in science journal Nature, SleepFM was trained between 1999 and 2024 using 580,000 hours of sleep data from 65,000 patients.

This data came from sleep clinics and medical facilities, and was split into five-second increments.

This allowed SleepFM to learn ‘the language of sleep’, James Zou, an associate professor of biomedical data science at Stanford, said.

To go alongside the sleep data, the researchers put forward individual health files on the sleep clinic patients.

Over the course of the study, it was right 80 percent of the time when predicting the likelihood of the following conditions – Parkinson’s, Alzheimer’s, dementia, hypertensive heart disease, heart attack, prostate cancer, and breast cancer.

It had a higher accuracy rate with a more morbid stat – death – in which it predicted accurately in 84 percent of cases

Where SleepFM had some difficulty was with predicting patients with chronic kidney disease, stroke, and arrhythmia (Irregular heartbeat), but was still accurate 78 percent of the time.

There were certain body signals that alerted SleepFM to trouble, like a brain that ‘looked asleep’ with a heart that ‘looked awake’.

While the results were certainly impressive, it’s worth noting the limitations.

For example, the study looked at people who already had suspected health problems, hence why they were participating in a sleep clinic trial.

This raises questions about whether SleepFM could predict health conditions among the general public.

Technology is developing at a rapid rate

There seems to be no limit to what technology can do these days, least of all with regard to artificial intelligence.

Recently, we reported on a website that can find every picture of you ever published on the internet.

For one Norwegian man, he used Grok to guide him to the hospital after his ruptured appendix was missed.

And it’s not just humans being subjected to scrutiny from artificial intelligence – it’s cars too.

Cars are being subjected to AI too, whether it’s for a ‘scarily’ good oil change or being destroyed at wrecking yards.

DISCOVER SBX CARS: The global premium car auction platform powered by Supercar Blondie

Ben Thompson is a Senior Content Writer at supercarblondie.com. Ben has more than four years experience as a qualified journalist, having graduated with a Multimedia Journalism degree from News Associates. Ben specializes in writing about Teslas, tech and celebrity car collections.