BiAffect

Background: Research by our group and others have demonstrated the feasibility of using mobile phone derived metadata to model mood and cognition. Given the effects of age and mood on cognitive performance, it was hypothesized that using such data a model could be built to predict chronological age and that differences between predicted age and actual age could be a marker of pathology.Methods: These data were collected via the ongoing BiAffect study. Participants complete the Mood Disorders Questionnaire (MDQ), a screening questionnaire for bipolar disorder, and self-reported their birth year. Data were split into training and validation sets. Features derived from the smartphone kinematics were used to train random forest regression models to predict age. Prediction errors were compared between participants screening positive and negative on the MDQ.Results: Three hundred forty-four participants had analyzable data of which 227 had positive screens for bipolar disorder and 117 had negative screens. The absolute prediction error tended to be lower for participants with positive screens (median 4.50 years) than those with negative screens (median 7.92 years) (W = 508, p = 0.0049). The raw prediction error tended to be lower for participants with negative screens (median = −5.95 years) than those with positive screens (median = 0.55 years) (W = 1,037, p= 0.037).Conclusions: The tendency to underestimate the chronological age of participants screening negative for bipolar disorder compared to those screening positive is consistent with the finding that bipolar disorder may be associated with brain changes that could reflect pathological aging. This interesting result could also reflect that those who screen negative for bipolar disorder and who engaged in the study were more likely to have higher premorbid functioning. This work demonstrates that age-related changes may be detected via a passive smartphone kinematics based digital biomarker.

Objective Ubiquitous technologies can be leveraged to construct ecologically relevant metrics that complement traditional psychological assessments. This study aims to determine the feasibility of smartphone-derived real-world keyboard metadata to serve as digital biomarkers of mood. Materials and Methods BiAffect, a real-world observation study based on a freely available iPhone app, allowed the unobtrusive collection of typing metadata through a custom virtual keyboard that replaces the default keyboard. User demographics and self-reports for depression severity (Patient Health Questionnaire-8) were also collected. Using >14 million keypresses from 250 users who reported demographic information and a subset of 147 users who additionally completed at least 1 Patient Health Questionnaire, we employed hierarchical growth curve mixed-effects models to capture the effects of mood, demographics, and time of day on keyboard metadata. Results We analyzed 86 541 typing sessions associated with a total of 543 Patient Health Questionnaires. Results showed that more severe depression relates to more variable typing speed (P?<?.001), shorter session duration (P?<?.001), and lower accuracy (P?<?.05). Additionally, typing speed and variability exhibit a diurnal pattern, being fastest and least variable at midday. Older users exhibit slower and more variable typing, as well as more pronounced slowing in the evening. The effects of aging and time of day did not impact the relationship of mood to typing variables and were recapitulated in the 250-user group. Conclusions Keystroke dynamics, unobtrusively collected in the real world, are significantly associated with mood despite diurnal patterns and effects of age, and thus could serve as a foundation for constructing digital biomarkers.