How you move your mouse or swipe your phone may reveal more about your mental health than you think. A new machine-learning system called MAILA can detect signs of depression, anxiety, and other conditions—just from the way you interact with a screen.
The research
In a preprint posted to arXiv in November 2025, researchers Veith Weilnhammer, Jefferson Ortega, and David Whitney at the University of California, Berkeley trained MAILA on 18,200 cursor and touchscreen recordings from 9,500 participants. Each recording was paired with self-reported mental health labels—totaling 1.3 million data points.
MAILA tracks mental states along 13 clinically relevant dimensions, including arousal, valence, and mood. It can follow circadian fluctuations (e.g., how mood changes from morning to night) and detect changes caused by experiments that alter someone's arousal or emotional state. At the group level, the model achieves “near-ceiling accuracy” at distinguishing different mental health conditions.
Importantly, MAILA captures information that goes beyond what people say about themselves. When combined with large language models (like ChatGPT), it can improve their ability to infer a user's mental health.
Why it matters
Traditional mental health assessment relies on self-report questionnaires or clinical interviews—both time-consuming and not always accurate. MAILA offers a passive, scalable alternative using devices people already use every day. It could help identify mental health struggles earlier and more objectively.
For cognitive researchers, this opens a new avenue for “digital phenotyping”—using everyday digital activity to understand psychological function without expensive lab equipment.
What you can do
Pay attention to your own screen habits. Notice if your mouse movements become slower or jerkier when you're feeling low—MAILA suggests these patterns are meaningful. While you can't run MAILA at home, you can stay curious about how your digital behavior reflects your mental state. For a broad cognitive snapshot, try a validated online assessment.
Source: arXiv q-bio.NC
Curious about your own brain? Take our free adaptive IQ test or try 306 brain training levels.
