Development of a Severity Score and Comparison With Validated Measures for Depression and Anxiety: Validation Study

Area of Science:

• Digital Health
• Behavioral Health
• Artificial Intelligence

Background:

• Measurement-based care (MBC) is underutilized in behavioral health, with less than 10% of patients receiving it.
• Mobile health (mHealth) platforms offer a potential solution for secure and efficient MBC implementation.
• An mHealth platform was developed to enhance MBC delivery and accessibility in integrated care settings.

Purpose of the Study:

• To evaluate a novel mHealth-derived severity score as a screening and monitoring tool for depression and anxiety.
• To compare the mHealth severity score against validated assessments (PHQ-9, GAD-7) and expert clinical review.
• To determine the added value of the mHealth severity score in behavioral health assessments.

Main Methods:

• An artificial intelligence (AI) algorithm was developed to generate a quantifiable severity score from patient-reported and passively collected mHealth data.
• The severity score was validated against the Personal Health Questionnaire-9 (PHQ-9) and Generalized Anxiety Disorder-7 (GAD-7) in over 35,000 users.
• Comparisons were made using linear regression, Bland-Altman analysis, and expert clinician review for a subset of 250 individuals.

Main Results:

• The mHealth severity score demonstrated strong correlations with changes in PHQ-9 (r=0.74) and GAD-7 (r=0.80) scores.
• A strong positive correlation was observed between the severity score and expert clinician review (r=0.80-0.84).
• Bland-Altman analysis indicated significant differences between the severity score and PHQ-9, suggesting unique information provided by the mHealth tool.

Conclusions:

• The mHealth severity score is a reliable proxy for longitudinal screening and monitoring of behavioral health symptoms.
• This novel measure may enhance the identification of at-risk individuals, potentially capturing those missed by the PHQ-9.
• Further research is recommended to fully establish the sensitivity and specificity of the mHealth severity score.