Sleep researchers have spent decades studying what happens inside the bedroom — mattresses, light, noise, temperature. A new study led by Professor Daisuke Matsushita at Osaka Metropolitan University, published in the journal Building and Environment, looks instead at what's outside the front door, and finds that the visual character of a person's street is measurably linked to how long and how well they sleep.
Machine Learning Meets Sleep Science
The research team analyzed more than 200,000 Google Street View images using machine learning to score visual characteristics of neighborhoods across Tokyo, including the amount of greenery, the degree of visual "enclosure" from surrounding buildings, and how walkable a street appeared based on features like sidewalks and traffic signage. Separately, the same AI models estimated how safe each streetscape would feel to a pedestrian, based purely on its visual appearance.
Those neighborhood scores were then linked to self-reported sleep duration and insomnia symptoms from 1,089 working adults living on lower floors of apartment buildings across the city, a population chosen specifically because ground-level and lower-floor residents are more exposed to their immediate street environment than upper-floor residents looking out over rooftops.
Greener and More Enclosed, Not More Walkable
Two neighborhood characteristics stood out. Streets with more greenery, such as tree-lined sidewalks, were associated with longer sleep duration. Streets with higher "enclosure" — meaning more surrounding buildings and fewer wide-open sightlines — were associated with both longer sleep and fewer insomnia symptoms.
A third finding ran counter to conventional urban-design wisdom: streets scored as more walkable, based on infrastructure like sidewalks and crosswalk signage, were associated with lower perceived safety and shorter sleep, not better outcomes. The researchers suggest that walkability infrastructure in isolation, without other reassuring visual cues, may not read as "safe" the way greenery and enclosure do, since it says nothing about how populated, overlooked, or monitored a street feels after dark.
Safety as the Mechanism
The study's more novel contribution is showing that greenery's effect on sleep wasn't entirely direct — it operated in part through perceived safety. Streets that scored as visually safer were associated with better sleep outcomes independent of other characteristics, and the AI-estimated safety scores helped explain why greener streets were linked to longer sleep in the first place. That distinction matters for urban planning: it suggests that simply adding trees for their own sake may be less effective than designing streetscapes where greenery also visibly signals safety, activity, and natural surveillance.
The study is cross-sectional, meaning it captured a single snapshot of both streetscape characteristics and self-reported sleep rather than tracking changes over time, so it can't establish that altering a street's appearance would directly improve a resident's sleep. Sleep was also self-reported rather than measured with wearables or polysomnography, and the sample was limited to lower-floor residents in one megacity, which may not generalize to suburban, rural, or different cultural contexts.
What This Means for Patients
For most people, moving isn't a realistic response to a study like this. Its value lies in reframing insomnia and short sleep as something more than a personal or bedroom-level problem: the visual environment just outside a person's window may be quietly working for or against their sleep every night. For clinicians and patients troubleshooting persistent sleep difficulty, it's a reminder that environmental factors worth asking about extend beyond light and noise in the bedroom to the streetscape a person walks through and glances at before bed. For city planners and housing developers, the findings offer a concrete design lever: neighborhoods that feel safe and enclosed, with visible greenery, may deliver a public health benefit as measurable as any single sleep hygiene tip, at a population scale no individual behavior change can match.
