Millions of people living with bipolar disorder use smartphone apps to monitor their mood on a daily basis, but a major new review has found that the technology still lacks the scientific rigour needed to be fully trusted as a clinical tool. Researchers say that whilst mood-monitoring apps hold genuine promise for improving mental health care, the evidence base remains fragmented and inconsistent.
The systematic review, published in the International Journal of Bipolar Disorders, analysed 42 studies involving more than 7,800 participants and examined how well both active and passive forms of digital mood monitoring performed over periods of at least three months. Active ambulatory assessment refers to tools that ask users to self-report their mood, sleep, or symptoms. Passive monitoring collects data automatically in the background, using smartphone sensors to track movement, phone usage, GPS location, and other behavioural signals without requiring any manual input.
The research team, led by Laurence Astill Wright at the University of Nottingham, found that some active monitoring approaches, particularly straightforward self-report questionnaires delivered via smartphone, performed reasonably well when compared against established clinical measures. However, the picture for passive monitoring was far less encouraging. Very few studies had examined how well passive data collection actually reflects mood states, and none demonstrated strong, consistent performance against recognised clinical scales.
One of the most striking findings was the sheer volume of variation across different studies. Across the 28 distinct ambulatory assessment systems examined, researchers identified 487 different smartphone-based performance measures. The measures differed so substantially in how they were defined, collected, and reported that it was impossible to pool the data through standard statistical methods. No two studies were measuring the same things in comparable ways.
Bipolar disorder presents a particular challenge for researchers because mood can shift dramatically over weeks and months. Conventional clinical assessments, which typically ask patients to reflect on the past week or month, may miss these rapid fluctuations. Digital tools that monitor behaviour continuously or in real time could, in theory, offer clinicians and patients a much richer picture of illness patterns.
The authors argue that passive monitoring technology, which draws on data such as step counts, heart rate, and communication patterns, is still poorly developed and insufficiently validated. They warn that without standardised reporting frameworks across research trials, progress in the field will remain slow and unreliable. They call on researchers to include validation data within their primary published results rather than in separate supplementary papers, which are often difficult to locate.
The review also notes that many people with bipolar disorder are enthusiastic about using digital tools to manage their condition, underscoring the importance of ensuring those tools actually work as intended. Trust in a mood-monitoring metric, the authors suggest, requires not perfection but sufficient accuracy to support meaningful clinical decisions.
Greater standardisation in how mental health apps are tested and reported is urgently needed before they can be widely adopted in psychiatric care.