2 - (SYM 111) Leveraging Patient-device Interaction Data from AKL-T03, a Digital Therapeutic That Targets Attentional Control in Depression, to Measure Real-time Cognitive Change

Background: A theoretical advantage of digital therapeutics (DTx) is the ability to provide objective, real-time assessment of patients during cognitive-behavioral treatments, without user input beyond engaging with the treatment itself. Randomized controlled trials (RCTs) support the efficacy of AKL-T03, a DTx that targets attentional control via a multitasking treatment game (navigation and targeting tasks), in depression; however, there remains a need to understand how DTx data may be used to monitor changes in cognition. We aimed to derive and validate a real-time measure of attentional control from AKL-T03 data in depression.

Methods: We conducted secondary analyses of STARS-MDD (NCT03310281), a 6-week RCT of AKL-T03 versus an active digital control in adults with major depressive disorder. Only the AKL-T03 group (n=37; M age = 43.11; 62% female) was included. A cognitive metric was derived from targeting response speed, targeting accuracy (d-prime), and navigation ability (80% psychometric threshold); a Bayesian hierarchical linear-log model estimated patient-level parameters. We regressed changes in the Test of Variables of Attention (TOVA)-Attention Comparison Score (ACS), an FDA-cleared neuropsychological measure of attention, on changes in the cognitive metric, controlling for TOVA-ACS baseline, cognitive metric baseline, and sex. We explored associations with additional cognitive outcomes and depression symptoms on the Patient Health Questionnaire-9 (PHQ-9).

Results: Most subjects showed a consistent logarithmic trend in the cognitive metric over treatment. Increases in the cognitive metric during treatment significantly predicted increases in TOVA-ACS (B = .120, p = .009; Figure 1), with a small effect size (Cohen’s d = .27). Exploratory analyses indicated that cognitive metric changes were also linked with changes in measures of working memory, visuospatial attention, and processing speed, but unrelated to change in PHQ-9. 

Discussion: Findings support the validity of a real-time attentional measure from AKL-T03 patient-device interactions. Our study was limited by a small sample; further, the non-significant association of the cognitive metric and PHQ-9 raises questions about the mechanisms by which AKL-T03 produces depression symptom change. Our findings underscore the potential of DTx for in-vivo monitoring of cognitive change in treatment. Given their low cost, scalability, and 24-hour availability, DTx may expand access to evidence-based assessment and treatment for underserved populations.