Smartphones Could Soon Predict Your Health Risks

The thought that the gadget you currently have on your nightstand—the one you picked up at two in the morning to check the time, the one you tapped through on your way to work in the morning—might know more about your health than your doctor does is subtly unsettling.

However, that’s about where we are. For years, researchers have been tugging at this thread, and the picture that is emerging on the other end is beginning to look serious.

A University of Michigan study that was published in JAMA Network Open tracked 557 adults for 15 days while allowing their phones to function normally. There are no buttons to push or surveys to complete. The sensors merely monitored people’s daily routines, such as when they got out of bed, how often they charged their phones, and when they moved. When the researchers combined those seemingly insignificant behaviors, they began to predict something more significant. Individuals who scored higher on what psychologists refer to as the “p-factor”—a broad measure of mental susceptibility—tended to spend more time at home, sleep in later, and leave their phones on low for longer periods of time.

It’s the kind of discovery that, when you hear it, seems obvious, but when you sit with it, it becomes unsettling. An uncharged phone. A walk in the morning was skipped. tiny signals that, by themselves, have no meaning. However, when stacked over days and weeks, patterns emerge that even a considerate friend might overlook.

The study’s lead psychologist, Prof. Aidan Wright, appears cautious about making too many claims. He notes that the DSM-5 is frequently an inaccurate tool because patients often have overlapping diagnoses and their behaviors seldom fit neatly into a single category. Passive data, on the other hand, provides something more akin to a continuous read. It’s more of a slow weather report of the mind than a diagnosis.

The uses go far beyond mental health. A heart rate can be detected by pressing a fingertip against a camera lens. Sleep apnea’s erratic pauses can be detected with a microphone placed by the bed. Of all things, speakers are being used to track breathing and reflect sonar off the chest. In a study conducted at the University of Illinois Urbana-Champaign, walking patterns and mortality risk were modeled using phone sensor data from 100,000 participants. The phone picks up when you walk a little less, a little slower. It doesn’t matter if you do or not.

It’s challenging to get medical equipment into homes and hospitals, but practically everyone already has a networked computer in their pocket, according to Dr. Andrew Gostine, who leads the sensor company Artisight. The percentage of Americans over 65 who own a smartphone has increased from a stubborn 13% ten years ago to over 60%. Something that was already in motion was accelerated by the pandemic.

However, there is still a disconnect between practice and potential. A promising but limited smartwatch app was developed by the University of Twente years ago to assist individuals with borderline personality disorder in identifying rising emotions through movement and heart rate data. Similar promise was shown by a relapse-prediction app for psychosis, as long as it was integrated into an actual treatment plan rather than operating independently.

It’s difficult not to feel a little tense as you watch this happen. The phone is both intimate and convenient at the same time. It probably depends more on who ends up with the data than on the technology whether that intimacy turns into care or surveillance. For now, that part is still genuinely unclear.