Collective Intelligence Increases Diagnostic Accuracy in a General Practice Setting.

Area of Science:

• Medical Decision Making
• Artificial Intelligence in Healthcare
• General Practice Research

Background:

• General practitioners (GPs) face diagnostic challenges in complex environments, leading to prevalent diagnostic errors.
• Previous research suggests aggregating independent diagnoses can enhance accuracy across various settings.
• The study investigates the potential of aggregating GP diagnoses to improve diagnostic accuracy.

Purpose of the Study:

• To determine if aggregating independent diagnoses improves diagnostic accuracy for general practitioners (GPs).
• To explore the combined benefit of diagnosis aggregation and a decision support system (DSS).

Main Methods:

• Simulated virtual GP groups using data from two prior studies (vignette-based and patient-facing).
• GPs provided diagnoses under control and/or DSS conditions.
• Virtual groups (3-9 GPs) used collective intelligence rules (plurality, confidence, seniority) to determine final diagnoses.

Main Results:

• Aggregating diagnoses using the plurality rule (equal weighting) significantly outperformed individual GP accuracy, with benefits increasing with group size.
• Diagnosis selection based on confidence showed minor improvements, while seniority-based selection reduced accuracy.
• Integrating the plurality rule with a DSS further enhanced diagnostic performance.

Conclusions:

• Aggregating independent GP diagnoses shows potential to substantially improve diagnostic accuracy and patient outcomes.
• The effectiveness of this collective intelligence approach varied across different patient cases.
• Future research should identify specific conditions where collective intelligence is most advantageous in general practice.