AI Analysis Of "Pokémon Sleep" Data Identifies Problematic Sleep Patterns

TOKYO, Feb 17 (Bernama-Kyodo) -- A leading Japanese sleep researcher has analysed the relationship between sleep and work performance using data collected through the "Pokemon Sleep" smartphone app, identifying sleep patterns that negatively impact productivity, Kyodo News reported.

Masashi Yanagisawa, a University of Tsukuba professor of neuroscience who supervised the app's development, used artificial intelligence to assist his team's analysis of a vast dataset on sleep quality, which was previously difficult to measure.

The game app enables users to raise and befriend Pokémon by recording their sleep time and detecting body movements to determine sleep and wake states when placed on the bed. It can also measure how long it takes the user to fall asleep, their sleep depth and how often they wake up during the night.

Yanagisawa and his university research team analysed around 2.1 million nights of sleep data, collected over 28 days from over 79,000 working adults in Japan who consented to participate in the study.

The five sleep phenotypes of healthy, long, fragmented, poor, and social jetlaggers -- characterized by misaligned sleep rhythms between weekdays and weekends -- were identified using AI to analyse sleep characteristics.

According to a paper released in December last year, when compared with survey responses about work efficiency, men and women in the latter two groups showed the worst scores for insomnia, daytime sleepiness, and presenteeism, defined as reduced work productivity despite being physically present at work.

The decline in productivity was most severe among social jet lag sufferers, with an estimated annual economic loss of about 140,000 yen (US$914) per person compared with those who have healthy sleep patterns. If the 16 per cent prevalence rate of social jet lag is extrapolated to Japan's entire workforce, the total annual loss is estimated to be about 1 trillion yen.

The study also found that productivity declines when sleep duration is too long or too short, leading the team to conclude that six to nine hours of sleep is ideal. They said that consistently long sleep durations may indicate underlying health issues, such as depression or sleep apnoea, and warrant caution.

The research team stressed that promoting regular sleep is crucial to preventing reduced workplace productivity. Since trying to make up for lost sleep on the weekends easily disrupts the body clock, Yanagisawa recommends going to bed 30 minutes earlier on weekdays instead.

Large-scale data collection had previously been a challenge, as traditional sleep research relied on self-reported sleep data from a relatively small number of participants.

Moving forward, the team hopes to use smartphones to support improvements in individual sleep habits and to help create workplace environments that balance health and productivity.

-- BERNAMA-KYODO