Sex and gender in digital mental health: why the data are not neutral.

Digital tools, including smartphone sensors, wearable devices, and conversational artificial intelligence, have moved mental health assessment beyond the clinic and into everyday life, offering the possibility of earlier detection and broader access. However, these benefits are not distributed equally between women and men. Drawing on recent evidence, this review traces sex and gender across four layers of the digital mental health pipeline: who reaches these tools, how algorithms interpret the data, how effectively interventions perform, and how raw sensor signals are understood. At each layer, biases that may appear to be technical details instead reflect sex- and gender-related patterns. Women are more willing to try digital mental health services but are less likely to complete them, largely because of time scarcity rather than low motivation. Diagnostic algorithms trained on male-skewed data underperform for women, sometimes markedly. The same sensor signal, such as a shrinking movement radius or a decline in heart-rate variability, can carry different meanings depending on the user's body and circumstances. These disadvantages can accumulate, leaving women doubly underserved. Yet the same evidence indicates that when sex and gender are incorporated into design, digital tools can reduce gaps rather than widen them. Sex and gender are not nuisance variables to be controlled away; they are parameters for precision care.