Usability-2

Formative Usability Testing
Conducted early in development to identify and resolve potential use-related issues through user feedback and observation.
Summative (Validation) Usability Testing
Conducted on final, market-ready devices to validate that intended users can operate the device safely and effectively, per IEC 62366-1 and FDA guidance.
Heuristic Evaluation
Expert review based on usability principles (e.g., visibility of system status, error prevention).
Task Analysis
Breaks down user workflows to identify high-risk or error-prone steps.

Explainability and Interpretability Testing
Evaluates whether the AI system explains its decisions in a way that is understandable and trustworthy to users (e.g., visual heatmaps, risk scores, confidence intervals).
Confidence Calibration Testing
Determines whether users appropriately trust or question AI outputs based on provided confidence levels or uncertainty measures.
Alert Fatigue / Notification Management Testing
Validates that AI-generated alerts or recommendations are clear, actionable, and not overwhelming.
Decision Support Acceptance Testing
Measures how likely users are to follow AI-generated recommendations and under what conditions they override or ignore them.
Bias Perception Testing
Evaluates user perception of fairness and reliability of AI outputs across diverse patient groups.

User Override and Manual Control Testing
Ensures that users can easily override AI actions or predictions when needed, with proper fail-safes in place.
Human-in-the-Loop Workflow Simulation
Tests how clinicians interact with the AI in real clinical environments or realistic scenarios (e.g., triaging patients, diagnostics).
Cognitive Load Testing
Measures mental workload and ease-of-use when interpreting AI outputs during real-world tasks.
Response Time and Feedback Loop Testing
Validates the responsiveness of the system to user inputs or corrections.

Graphical User Interface (GUI) Usability Testing
Tests design clarity, intuitive layout, navigation, and readability of AI dashboards or decision-support tools.
Voice/Touch Interaction Testing
Especially important for wearable AI devices, robots, or smart assistants used in surgical or patient-care settings.
Accessibility Testing
Ensures the interface accommodates users with visual, motor, or cognitive impairments (per WCAG, Section 508, or EN 301 549 standards).

Use Error Analysis
Identifies and categorizes user errors and their root causes (especially critical for AI systems where incorrect interpretation may be subtle).
Use-Related Hazard Identification (linked to ISO 14971)
Maps device use scenarios to potential hazards caused by interaction with AI elements (e.g., incorrect diagnosis suggestions).
Residual Risk Acceptability Testing
Assesses user tolerance and system responses to residual use-related risks.

Persona-Based Testing
Involves multiple representative user types (e.g., nurse vs. physician, novice vs. expert) to validate usability across all target audiences.
Scenario-Based Usability Testing
Simulates real clinical scenarios using the AI system, with observation and feedback.
Think-Aloud Protocol
Users narrate their thought process while using the AI tool—helpful for identifying friction or confusion in real-time.

Usability Engineering File (UEF)
Complete documentation of the usability engineering process (per IEC 62366-1).
Human Factors Summary Report (for FDA submissions)
Includes test plans, results, analysis, and justifications that support safety and effectiveness.
AI-Specific Usability Risk Report
Documents unique usability risks related to algorithm transparency, bias, and explainability.