Clinical Scorecard: Who Decides How AI Enters Pathology?

At a Glance

Key Highlights

• Google AI's Lymph Node Assistant (LYNA) achieved high accuracy in detecting metastatic breast cancer.
• Clinical validation in real-world settings is essential before routine use of AI tools.

Guideline-Based Recommendations

Diagnosis

• AI tools should be validated in the local clinical environment before patient care.

Management

• Implementation should start with a clear clinical need.

Monitoring & Follow-up

• Ongoing monitoring for version changes or drift is necessary.

Risks

• Human oversight is important to define when AI can be relied upon.

Patient & Prescribing Data

AI can support diagnostic tasks.

Clinical Best Practices

• Ensure accurate annotation and establish consensus-based 'ground truth' for datasets.
• Promote diverse and representative datasets to reduce bias.
• Define expectations for explainability.

Related Resources & Content