Predicting Therapy Success and Costs for Personalized Treatment Recommendations Using Baseline Characteristics:...

Area of Science:

• Psychiatry and Mental Health
• Health Economics
• Artificial Intelligence in Healthcare

Background:

• Clinical and cost-effectiveness are vital for healthcare decision-making in psychological disorders.
• Identifying which patients benefit most from specific interventions and understanding individual costs is challenging at the outset of treatment.

Purpose of the Study:

• To predict individual patient outcomes and treatment costs before initiating internet-based interventions for psychological disorders.
• To develop a framework for personalized treatment recommendations based on predicted outcomes and costs.

Main Methods:

• Utilized machine learning techniques on baseline data from 350 patients in a randomized controlled trial for depressive disorders.
• Evaluated and compared the predictive accuracy of various machine learning models.
• Integrated predictions with incremental cost-effectiveness analysis to generate treatment recommendations.

Main Results:

• Machine learning predictions of clinical outcomes and costs surpassed a predefined reference measure.
• Patient-reported anxiety, depression, mobility, and energy levels were key predictors.
• Personalized recommendations, at a threshold of €25,000/QALY, suggested slightly worse outcomes (1.98%) but reduced costs (5.42%).

Conclusions:

• Personalized treatment recommendations at baseline are feasible for psychological disorders.
• This approach can enhance healthcare decision-making, leading to better individual outcomes and reduced costs.