Understanding data collection behaviour of mental health practitioners

Area of Science:

• Clinical Informatics
• Artificial Intelligence in Healthcare
• Mental Health Research

Background:

• Suicide risk assessment is critical for patient safety.
• Clinical decision support systems (CDSS) aim to improve accuracy.
• GRiST CDSS faces challenges with variable data collection.

Purpose of the Study:

• To explore clinician data collection behavior within the GRiST CDSS.
• To identify important variables for suicide risk evaluation.
• To develop a dynamic data collection model.

Main Methods:

• Analysis of clinician data collection patterns with GRiST.
• Investigating relationships between higher-level 'branch' questions and data relevance.
• Utilizing GRiST's hierarchical knowledge structure.

Main Results:

• Clinician data collection varies significantly based on patient profiles and assessment context.
• Higher-level questions can guide assessors to critical data points.
• A model for dynamic, knowledge-hierarchy-driven data collection was developed.

Conclusions:

• Dynamic data collection driven by knowledge hierarchy can optimize GRiST.
• This approach has potential for other CDSS with high-dimensional, partially collected data.
• Improved data collection can enhance suicide risk assessment accuracy.