AI for Healthcare Capstone Exercise — AI for Healthcare Workers

What you'll learn

Apply skills from across this course in a single realistic professional scenario
Produce a concrete, role-relevant deliverable using AI tools
Self-assess your output against professional quality criteria

This course has covered the clinical, regulatory, and ethical dimensions of AI in healthcare: documentation tools, diagnostic support, patient data privacy, professional accountability, administration, patient communication, and clinical research. The capstone places you in a governance scenario that brings nearly all of those dimensions together.

Evaluating AI tools for clinical adoption is one of the most consequential tasks a senior clinician can take on. The evaluation framework you produce will be read by people who are not AI experts. It needs to be structured enough to enable consistent scoring, honest about where clinical judgment cannot be replaced by a rubric, and grounded in the regulatory and safety requirements that apply to your setting.

Capstone Exercise

Clinical Evaluation Framework for Ambient Scribing Tools

Context

You are a senior nurse in an outpatient clinic at an NHS trust. Your manager has asked you to evaluate two AI-powered ambient scribing tools that are being considered for a pilot. The evaluation needs to be submitted to the clinical governance committee in one week. The committee includes the medical director, the information governance lead, and two consultant physicians. None of them will have time to read a long report. They need a structured framework that covers capability, safety, regulatory compliance, and workflow impact, plus a scoring rubric they can use to compare the two tools after the pilot.

Your Task

Using an AI assistant such as Claude or ChatGPT to support the research and writing process, draft a clinical evaluation framework document of 500 to 650 words. The document must contain five clearly labelled sections: (1) Key capability criteria the tools must meet to be clinically useful; (2) Patient safety considerations including failure modes and escalation requirements; (3) Regulatory and data governance requirements covering UK GDPR, NHS data standards, and clinical risk classification; (4) Staff workflow impact covering documentation burden, training requirements, and clinician override capability; (5) A scoring rubric in table format that the committee can use to compare the two tools across the criteria in sections 1 to 4. Below the document, add a brief note of three to five sentences identifying which sections required clinical judgment that AI could not supply, and how you addressed that in the drafting process.

Your notes (optional)

Deliverable

A clinical evaluation framework document of 500 to 650 words with five labelled sections and a scoring rubric table, followed by a three-to-five-sentence note identifying the sections that required clinical judgment AI could not supply. The document should be ready to submit to a clinical governance committee without further structural changes.

Quick check

The capstone insists the one-to-five scoring rubric carry anchor descriptors at 1, 3 and 5. What problem do those anchors solve for the governance committee?

Select one answer.

Key takeaways

AI can accelerate the structural drafting of clinical evaluation documents but cannot supply the clinical judgment required to define safety criteria, identify failure modes, or validate regulatory classification accurately
Governance-facing documents must identify where AI was used in their preparation and where human clinical expertise overrode or replaced AI output: this is both a professional accountability standard and an emerging expectation in NHS settings
Scoring rubrics are only useful if they have anchor descriptors: a number without a behavioural description tells a committee nothing about what a 3 versus a 5 looks like in practice
The most valuable skill this exercise tests is knowing which sections to brief the AI on carefully, which to draft yourself and have AI refine, and which to write entirely from clinical knowledge

Clinical Evaluation Framework for Ambient Scribing Tools

Try these prompts