Skip to content

Evaluators

This section documents the individual evaluators used in ASHAI's medical response assessment system.

Overview

Each evaluator focuses on a specific dimension of medical AI response quality:

Composite Scoring

The composite evaluator combines all individual evaluators with the following weights:

  • Medical Accuracy: 45% - Most important for medical safety
  • Precision: 25% - Directness and relevance
  • Language Clarity: 20% - Communication quality
  • Empathy: 10% - Patient-centered communication

Integration

All evaluators are integrated with Weave for comprehensive trace analysis and leaderboard management.

Generated Evaluators

Generated Evaluators