Receiving Information on Machine Learning-Based Clinical Decision Support Systems in Psychiatric Services Increases...

Area of Science:

• Psychiatry
• Health Informatics
• Artificial Intelligence

Background:

• Machine learning (ML) models are increasingly used in clinical decision support systems (CDSS) within psychiatry.
• Healthcare staff trust is crucial for the successful implementation of these ML-based CDSS.
• This study investigates the impact of providing information about ML-based CDSS on staff trust.

Purpose of the Study:

• To determine if educating healthcare staff about ML-based CDSS enhances their trust in these systems.
• To assess the effect of information provision on both trust and distrust levels.

Main Methods:

• A randomized survey experiment was conducted with psychiatric healthcare staff.
• Participants were assigned to three groups: intervention (ML information), active control (standard process information), or blank control (no information).
• Trust and distrust scores were compared between groups using pairwise comparisons.

Main Results:

• The intervention group showed significantly enhanced trust and diminished distrust compared to both active and blank control groups.
• Trust increased by 5% in the intervention arm versus the active control, and distrust decreased by 4%.
• Statistical significance was maintained across comparisons (p < 0.05).

Conclusions:

• Providing basic information about ML-based CDSS can effectively increase healthcare staff trust in these systems within hospital psychiatry.
• This finding supports strategies for improving the adoption of AI-driven tools in mental healthcare settings.