HEALTH

Phone Sensors May Measure Minds Differently

United StatesTue Nov 17 2026

Phone sensors may someday help screen mental health problems. They watch what people do. They do not know how people feel. They can act in different ways. That is important before any tool is used for symptom checks or tracking.

A team looked at data from four US college groups. Those groups were from 2018 to 2021. The study had 705 yearly records. It had 497 people. The data held 554 iPhone records and 151 Android records. The team did not link people across years. So it ran separate tests for each group.

The team compared 15 raw measures and matched changes. It used Cliff delta, resampling, and fixed logistic regression. It split phone location and screen data from Fitbit activity and sleep data. It checked coverage, complete cases, psychological adjustment, and past depressive symptoms. It used the prior 14 days.

The team found lower Bluetooth counts on iPhone in every group. The Cliff delta was -1.00 to -0.85. Connectivity counts showed larger platform gaps than core phone behavior. The mean absolute Cliff delta was 0.22 to 0.43. All 95 percent confidence intervals were above zero. Small value gaps did not mean equal coverage. In 2020, 41 of 103 iPhone records lacked all home-time values. That was zero of 34 Android records. Platform classification AUROC was 0.94 to 0.997 for 13 raw measures.

actions