IoT and Sleep Medicine:

changing paradigms, changing lives moving treatment from the hospital to the home

Dr. Wen-Te Liu

  • Doctor, Department of Chest Medicine, Taipei Medical University-Shuang Ho Hospital, Ministry of Health and Welfare.
  • Associate Professor, Professional Master Program in Artificial Intelligence in Medicine, College of Medicine.
  • Assistant Professor, Department of Internal Medicine, School of Medicine Associate Professor, School of Respiratory Therapy

Find more about his research

If you are a snorer, tired during the day, hoarse in the morning, or have to wake up at night to visit the bathroom, you might be suffering from sleep apnea. It is a common problem – perhaps half of patients seeing a GP will have some level of sleep apnea – that can have serious complications. Sleep apnea is correlated with elevated risk of heart disease and stroke, and has recently been linked to Alzheimer’s disease.

The good news is the condition is usually treatable with lifestyle modifications or a ventilator, but hospital based testing is expensive and time consuming, and according to research by Dr. Wen-Te Liu, traditional hospital-based diagnosis may not be as accurate as previously thought.

The problem of accurately and efficiently measuring the factors that affect sleep and creating effective treatment programs is TMU’s sleep medicine expert is looking to address, and to do it he is making use of booming developments in internet-of-things (IoT) devices and artificial intelligence.

Sleep apnea is a dysfunction of the autonomic nervous system where muscles in the back of the throat relax and block the airway, temporarily stopping breathing. When the brain detects a low blood oxygen level, it briefly wakes you up to kick-start breathing. The process is often too quick to notice, and that can happen dozens of times each hour.

Traditional sleep medicine diagnosis involves doctors subjecting patients to a battery of questions before sending them for a night in the sleep lab. Researchers previously expected that sleep apnea would be underestimated in the lab due to sleeping in a strange environment connected to wires. But research published by Dr. Liu last year actually showed that the so called “first night” effect may actually result in the condition being overestimated in the lab, probably because patients connected to so many wires that they are forced to sleep lying on their backs and can’t change their positions, which increases sleep apnea severity.

Dr. Liu wanted to predict sleep apnea severity without relying on hospital based tests. He began by gathering body profile and other medical parameters and came up with a prediction model that was 70-80% accurate. “Maybe it’s because the patients’ condition varies in the hospital, so I thought we need to test patients’ sleep apnea in their home,” said Dr. Liu.

Advances in artificial intelligence tools and the IoT now make home-based measurements possible. Dr. Liu is now undertaking a four-year study project that incorporates lab and home based measurements of patients’ sleep quality using wearables and IoT sensing devices to create a comprehensive sleep apnea prediction model that allows individualized precision treatment strategies.

The first step is to streamline preliminary information gathering. Instead traditionally spending up to 90% of the time asking patients questions, the preliminary questionnaires can be automated using an interactive chatbot program on a web- or mobile-based app.

In the second phase, the chatbot can gather data on voice quality and cognitive function. As a snorer himself, Dr. Liu felt hoarse after a poor night’s sleep, and he also saw many patients with persistent cough and hoarseness caused by snoring which led to the idea that voice could be as an indicator of sleep disturbance. The patient’s voice can be recorded using a cellphone app, and the AI system can determine neurocognitive function at the same time.

Voice and neurocognitive data combined with other parameters can be recorded using wearables and home-based IoT devices that gather physiological and environmental data to account for factors that affect sleep quality like physical activity, air quality, noise, light, and temperature. In the third phase, all the information is integrated and analyzed using AI tools to help create cognitive behavior therapy programs and sleep environment interventions.

In designing his model Dr. Liu is taking advantage of its access to data from over 10 000 patients in northern Taiwan. He has already successfully uncovered links between physical profiles, environmental pollution, and sleep disorders. As his AI model takes shape, a comprehensive sleep disorder prediction model is on the horizon.

Dr. Liu says that as sensing technology and data processing tools improve, doctors will be freed up to spend more time on patients, whether they have complicated clinical problems or would benefit from minor lifestyle changes. “When we developed some system or machine to take over the ‘low level’ task … we can find how to improve patients’ and peoples’ health. In 200 years of medical science, we just have the resources to focus on the severe problems.”

Besides providing patients with longitudinal follow-up, the smart devices used to monitor severe sleep apnea can also be useful for people with subclinical issues. Dr. Liu is also working with business to build a personalized care program for anyone to upload their own data for analysis by his AI model.

A large proportion of sleep disturbances can be addressed through lifestyle changes, but whether a patient’s condition can be improved by sleep hygiene or a sleep apnea ventilator, sleep data gathered in Dr. Liu’s program will be a valuable resource for research and precision treatment, and will help move sleep studies to where sleep actually happens – out of the hospital and into the bedroom.

For interviews or a copy of the paper, contact Office of Global Engagement via global.initiatives@tmu.edu.tw.