AI Prioritization Workshop Kit

Purpose and outcome

This one-day workshop helps clinical and operational teams identify, evaluate, and prioritize AI pilot opportunities that deliver measurable clinical or operational value while managing safety and implementation risk. Participants will leave with scored use-case candidates, a prioritized short list, and draft pilot charters for the top opportunities.

Who should attend

• Clinical lead(s) (physician, nurse, therapist) familiar with the workflow
• Operational lead (manager or service owner)
• Data/IT representative (data access, integration constraints)
• Quality & safety representative
• Patient experience or patient advocate (when relevant)
• Project or improvement facilitator

Preparation (recommended pre-work)

• Collect 6–12 candidate use cases (brief one-paragraph descriptions).
• Bring a simple data inventory: available data sources, owners, typical lags, and known quality issues.
• Identify relevant regulatory/privacy constraints (e.g., PHI, data sharing limits).
• Assign participants to come prepared to discuss clinical/operational impact and workflow change implications.

One-day agenda (flexible)

1. 0:00–0:20 — Welcome & framing

   Set goals, scope, and success criteria for the day. Agree the definition of an AI “pilot” (timebox, success metrics, safety checks).

2. 0:20–0:50 — Short primers

   Quick overview: what AI can/can’t do, common deployment patterns, safety considerations, and data realities.

3. 0:50–1:40 — Use-case presentations

   Each use-case owner presents a 5–7 minute snapshot: problem, expected benefit, users, necessary data, and known risks.

4. 1:40–2:10 — Clarifying Q&A & grouping

   Group similar use cases; remove duplicates; decide the 6–8 candidates you will score.

5. 2:10–2:40 — Scoring rubric introduction

   Explain criteria, scale, weighting, and scoring method. Do an example scoring together.

6. 2:40–3:20 — Small-group scoring

   Break into small teams. Each team scores 2–3 use cases using the worksheet and rubric.

7. 3:20–3:40 — Break & consolidation
8. 3:40–4:10 — Review scores & discuss differences

   Bring teams together, reconcile wide score variations, capture assumptions behind high/low scores.

9. 4:10–4:40 — Prioritization & decision rules

   Apply decision rules (thresholds, balance of clinical vs operational value, safety screens) and produce prioritized short list.

10. 4:40–5:00 — Draft pilot charters & next steps

    Assign owners and timelines for completing pilot charters and governance approvals.

Scoring rubric (use this to evaluate each candidate)

Use a 1–5 scale (1 = low / unlikely, 5 = high / very likely). Score each criterion and compute a weighted total.

Criteria & guidance

• Impact (weight 35%) — Expected clinical or operational benefit if successful. Consider patient outcomes, safety improvement, reduced LOS, provider time saved, or throughput gains.
• Feasibility (weight 25%) — Likelihood the pilot can be built and integrated within the timeframe. Consider technical complexity, team capacity, and vendor availability.
• Safety & clinical risk (weight 20%) — Patient safety implications and the degree of human oversight required. Lower safety risk increases priority; high-risk items require additional mitigation before piloting.
• Data readiness (weight 15%) — Availability, quality, and timeliness of the data needed to train/validate the model.
• Operational adoption (weight 5%) — Likelihood clinicians/staff will adopt the tool; consider workflow disruption and change management needs.

Example: Weighted score = 0.35*Impact + 0.25*Feasibility + 0.20*(5 - SafetyRisk) + 0.15*DataReadiness + 0.05*Adoption. (Invert safety risk so higher is better.)

Use-case worksheet (one per candidate)

Stakeholder map (quick template)

Map stakeholders by influence and interest. For each high-priority use case, list:

• Executive sponsor
• Clinical champion
• Data owner
• IT/integration contact
• Quality & safety reviewer
• End-user representatives

Pilot charter template (deliverable)

Use the charter to request approvals and to control scope.

1. Title and sponsor
2. Problem statement & objectives
3. Success metrics (primary + secondary)
4. Scope (in/out)
5. Data sources & access plan
6. Team and roles
7. Delivery timeline and milestones
8. Validation & safety plan (including human-in-loop, rollback criteria)
9. Deployment & monitoring plan (how performance and safety are measured post-launch)
10. Estimated resources and budget
11. Decision gate criteria (what counts as success to expand versus stop)

Decision rules & governance

Examples of pragmatic decision rules to use after scoring:

• Auto-approve pilots with weighted score >= 4.2 and Safety Risk inverted score >= 4.
• Require safety mitigation plan before piloting if Safety Risk <= 3, even if overall score is high.
• Prioritize at least one low-risk, high-feasibility pilot in the first wave to build trust and capability.
• Assign a 3-month validation window and require pre-defined metrics and a Go/No-Go review at the end of the window.

Facilitation tips & common pitfalls

• Start with a shared definition of success (clinical outcome + adoption metric).
• Force clarity on what is being automated vs. augmented — avoid vague promises of “automation”.
• Be explicit about safety: identify worst-case failure modes and required human oversight up-front.
• Capture assumptions and unknowns in the worksheet — these drive feasibility workstreams after the workshop.
• Don’t let novelty win: prioritize measurable return and realistic implementation paths over technological glamour.

Pre-workshop checklist (quick)

• Collect use-case briefs (6–12)
• Data inventory summary available
• Invite right stakeholders and confirm attendance
• Prepare blank worksheets and scoring sheets for teams
• Reserve a whiteboard or collaborative digital board for grouping and scoring

Post-workshop deliverables & suggested next steps

• Scored use-case register (spreadsheet) with assumptions and owners
• Draft pilot charters for top 2–3 candidates
• Data access & ethics checklist started
• Governance / review schedule for pilot approvals and safety reviews

Example short use-case (illustrative)

Use-case: Early detection of sepsis risk on admission. Problem: delayed recognition increases morbidity. Expected benefit: reduce time-to-antibiotics; measure as % reduction in time-to-antibiotics and sepsis-related ICU transfers. Data: vitals, labs, nursing notes, order history. Key risks: false positives causing unnecessary treatment; require human validation and conservative alerting threshold for pilot.

How this ties to our domain hungers

This kit helps healthcare teams choose pilots that improve patient safety, outcomes, and operational efficiency without creating unmanaged clinical risk. It balances eagerness to experiment with practical controls that preserve patient safety and stakeholder trust.

Capability enhancement opportunities

To make the workshop repeatable and easier to operate, consider:

• Converting the scoring rubric and use-case worksheet into an interactive form that saves submissions (so scores and assumptions are retained for audit and iteration).
• Adding a dashboard that aggregates scores across departments and displays trends in pilot outcomes.
• Packaging this kit as a reusable domain collection so sites can copy and tailor local scoring weights, safety gates, and pilot-charter templates.

Facilitator note: Keep the day pragmatic, document assumptions, and insist on a safety-first pilot design. The goal is not to pick the perfect project, but to select a small set of pilots with clear owners, measurable success criteria, and acceptable risk controls.