Clinical Scorecard: Who's Running QC on Artificial Intelligence?

At a Glance

Key Highlights

• AI tools are being deployed without a coherent governance framework.
• Clinical AI tools require rigorous validation and ongoing performance monitoring.
• Model drift can lead to silent degradation of AI tool performance.
• The laboratory medicine framework should guide AI governance.
• Current regulatory guidance for AI in medicine is lagging.

Guideline-Based Recommendations

Diagnosis

• Establish analytical validation processes before AI tool deployment.
• Define conditions for reliable AI tool performance.

Management

• Implement ongoing quality control measures for AI tools.
• Define action thresholds for performance review or removal.

Monitoring & Follow-up

• Monitor AI tool performance regularly and investigate shifts.
• Ensure visibility into AI validation and performance for clinicians.

Risks

• Accountability for underperformance must be clearly defined.
• AI tools can produce authoritative but incorrect outputs.

Patient & Prescribing Data

Patients receiving care influenced by AI outputs.

AI tools can enhance decision-making but require rigorous oversight.

Clinical Best Practices

• Incorporate laboratory medicine principles into AI governance.
• Engage pathologists and laboratory professionals in AI oversight.
• Ensure transparency in AI tool validation and performance metrics.

References

• Clinical AI Governance Framework