BiAffect

Functionality & Mechanism

BiAffect is a research application designed for passive digital phenotyping of mood states. The system leverages a custom virtual keyboard that replaces the default interface to unobtrusively collect keystroke dynamics metadata, including typing speed, error rate, and session duration. This data, along with passive sensor information from the device's accelerometer, is captured during routine use. Participants may also complete periodic self-report questionnaires to correlate objective behavioral data with subjective mood states, facilitating longitudinal analysis.

Evidence & Research Context

• In an analysis of over 86,000 typing sessions from 147 users, more severe self-reported depression was significantly associated with more variable typing speed (P<.001), shorter session duration (P<.001), and lower accuracy (P<.05).
• A feasibility study in a cohort with bipolar disorder demonstrated that passively collected metadata could be used to create statistically significant models predicting clinician-rated depression (HDRS, conditional R²=.63) and mania (YMRS, R²=.34) scores.
• The platform's kinematic data has been used to develop models that predict chronological age, finding significant differences in prediction error between participants screening positive versus negative for bipolar disorder risk.
• The association between keystroke dynamics and mood was found to persist even after statistically controlling for confounding variables such as diurnal patterns and the effects of aging on typing performance.

Intended Use & Scope

This application is intended for researchers investigating digital biomarkers for mood disorders. Its primary utility is as a platform for passive data collection to explore correlations between keyboard dynamics and mental health states. The system does not provide diagnostic outputs or clinical guidance and is not a substitute for professional medical evaluation or treatment.