Could your phone notice a stressful day before you do?

Think about how you use your phone on an ordinary day. You reply to a message, look something up, pay for something, close an email you’ll deal with later. None of it feels like information. It just feels like a day.

A team at KAIST, the Korea Advanced Institute of Science and Technology, wondered whether those small, repeated routines might quietly carry a signal. Not what you say, and not which app you open, but the shape of what you do. Their idea was that a stressful day might leave a faint pattern in how we move through our phones.

What the researchers did

The work, led by Hansoo Lee and Uichin Lee, was published in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies and is being presented at the UbiComp/ISWC 2026 conference. It was written up by News-Medical this week from a KAIST press release.

Instead of counting minutes spent in each app, which is the usual approach, the team looked for repeated behaviours: things like chatting in a messenger, browsing the web, making a payment, or writing an email. A language model helped label the patterns, and the system recorded things like how long and how often each behaviour happened, and whether text was typed. According to the researchers, the content of messages wasn’t stored, and information about who people contacted was scrambled.

Twenty-six people took part: university students and office workers in their twenties and thirties, all using Android phones. Some contributed around a month of data, others more than three months. Each day they also rated their own stress, and those ratings were what the models tried to predict.

The result was a personalised model for each person that could tell a stressful day from a calmer one around three times out of four. The best accuracy reported was about 76 per cent for the shorter-term group and about 74 per cent for the longer-term group. That was a few percentage points better than the conventional app-usage approach. The researchers also reported that messaging alone, or messaging plus payments, still did better than that conventional method.

The honest caveats

This is a small study, and it’s an early one. Twenty-six people is a starting point, not a verdict, and everyone involved was young and on the same type of phone. The authors themselves point out that the findings show an association between behaviour and stress, not a cause, and they plan to test other ages, occupations and devices. Some details, like which stress questionnaire was used and how the comparison was tested statistically, weren’t described in the press coverage we read. We haven’t read the full paper, so if the detail matters to you, it’s the place to look.

“Three out of four” also means about one in four days the model got it wrong. A pattern in the data isn’t the same as knowing how someone actually feels, and it doesn’t tell anyone what to do about a hard day.

Nagi’s take

What stays with us isn’t the technology. It’s the quiet idea underneath it: that stress shows up in small, ordinary habits long before we put a name to it. Anyone who has caught themselves rereading the same message four times, or putting off the easy thing, will recognise that.

Where we land is a gentler place. If the pattern is there, you are the person best placed to notice it, and you can do it on your own terms, without anything watching you do it.

That’s the thinking behind Nagi’s four-tap check-in. Mood, sleep, energy, a few seconds, and that’s all. It’s something you choose to do, not something that observes you. There’s no name and no email, and nothing to explain to anyone. Over a few weeks, some people find that seeing their own days laid out helps them spot what tends to go with what: the tense weeks and the restless nights, perhaps, or the days that were simply fine.

Nagi can’t tell you what any of it means, and it doesn’t diagnose or treat anything. It’s a calm space that’s just yours, and it holds the record steady so you can look.

A grounded close

If today has felt heavier than usual, you don’t need a model to tell you so. Your own sense of it counts. A slower hour, a glass of water, a message to someone who makes things feel lighter: small things are allowed to be enough.

And if stress has been sitting with you for a while, please don’t carry it alone. A friend, a family member, your GP or another professional you trust can help in ways an app never will.

If you want a private place to start noticing your own patterns, Nagi is on Google Play and the App Store — or in your browser at app.nagiandme.app. The first 15 messages are free. No name, no email. Just you, checking in.

— The Nagi team

Source: Lee, H., Park, T., Koh, Y., Lee, J.-G., & Lee, U. Leveraging smartphone human interaction routine behavior task mining and modeling for daily stress monitoring. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, December 2025. DOI: 10.1145/3770644. Reported by News-Medical, 14 October 2026 (KAIST press release). We read the news report only, not the full paper.

Nagi & Me is a mental health companion app, not a medical device, and nothing here is medical advice. If you’re struggling, please reach out to a professional or someone you trust.