A framework for precision "dosing" of mental healthcare services: algorithm development and clinical pilot

Area of Science:

• Computational psychiatry
• Digital health interventions
• Clinical decision support systems

Background:

• High prevalence of mental illness in the US, with over half of adults not receiving treatment.
• Lack of innovation in timely and appropriate mental healthcare service delivery.
• Need for personalized treatment planning to address individual patient needs.

Purpose of the Study:

• To develop and evaluate a computational framework for estimating the impact of mental healthcare services on depressive symptom severity.
• To assess the feasibility of using this framework for optimizing treatment schedules and supporting clinical decision-making.
• To pilot a dashboard prototype for clinicians to aid in treatment planning.

Main Methods:

• Leveraged historical observational clinical data from a virtual healthcare system.
• Developed a prototype computational framework conceptualizing mental healthcare as therapeutic interventions ('session dosing').
• Modeled 497 patients with severe depression symptoms and conducted a 5-week pilot with 22 providers and 126 patients using a dashboard.

Main Results:

• The framework successfully modeled patient symptom fluctuations, with 77% of the dataset fitting criteria for individual planning.
• 88% of individuals were identified as adequate for session optimization planning via the dashboard.
• In the pilot, 90% of clinicians used the dashboard, enabling discussions and identifying opportunities for automated session recommendations.

Conclusions:

• Modeling mental healthcare services can identify their potential to resolve depressive symptom severity fluctuations.
• Implementing this prototype framework advances mental healthcare treatment planning in real-world clinical settings.
• Further research is needed to assess clinical endpoint impacts and optimal integration into workflows.