Outcome Measurement in a Personalized, AI-Generated Digital Wellness Intervention: Instrument Selection, Administration Protocol, and a Rationale for Omitting Suicidal-Ideation Screening
Year: 2026
Type: report
Abstract: Background. Digital wellness applications are numerous and inexpensive to produce, but relatively few describe how they measure change in the people who use them, and fewer still publish that method in enough detail to be evaluated critically. Where measurement is described, the instruments and their administration schedule are often reported without a corresponding account of what happens when an instrument surfaces acute risk. Objective. To document the outcome-measurement methodology of a personalized, AI-generated audio intervention for sleep, stress, anxiety and habit change, and to set out the reasoning behind a specific design decision: administering the PHQ-8 rather than the PHQ-9, and therefore not screening for suicidal ideation. Method. The application administers the PHQ-8 and GAD-7 at baseline and on a 14-day re-administration cycle, without modification to item wording or response scale, alongside behavioural engagement measures. Aggregate reporting is subject to a minimum cohort size. Cost-avoidance is modelled observationally against published third-party estimates rather than measured directly. Rationale for instrument selection. The PHQ-9's ninth item screens for suicidal ideation. The platform employs no clinicians and operates no crisis service; a disclosure made through the application would reach no one qualified to act on it. The PHQ-8 is a separately validated instrument developed for population research in precisely this circumstance, correlates with the PHQ-9 at r = 0.998, and retains the ≥10 threshold for probable major depression. Crisis resources are instead presented on a permanently available, unauthenticated page rather than as a response triggered by a positive screen. Status. This is a methodology report. At the time of writing no cohort has accumulated sufficient enrolment to report aggregate outcomes, and none are presented here.
Author: Bull, Mark
Language: English
Subfield: Applied Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7204093942