Healthcare Innovation + AI
Healthcare Innovation and AI, read from the operational floor
My preparation includes AI in Healthcare: Stanford University. I approach this subject as an operator: interested in what technology changes about the work, and honest about what it does not.
Perspective
Fluency, not evangelism
Most conversations about AI in healthcare happen at two unhelpful extremes: uncritical enthusiasm, or reflexive dismissal. Neither is useful to a leader who has to decide whether a tool belongs in a workflow next quarter.
My interest is in the middle: building enough genuine fluency to evaluate claims, understand what a model can and cannot be responsible for, and design the governance that has to sit around it. Where I have applied technology directly, it has been in operational visibility — dashboards that put census, staffing, quality, and compliance in front of leadership in real time rather than in retrospect.
Preparation
AI in Healthcare: Stanford University. DNP Candidate. MBA-trained in cost and operations. Clinical and executive practice across business and healthcare.
Governance
Four questions before any tool touches care
What problem is this actually solving?
Technology introduced without a clearly named operational problem tends to add work rather than remove it. The first question is always whether the problem is real and worth the disruption.
Who is accountable for the output?
When a system contributes to a clinical or operational decision, accountability must remain with a named human role. Governance starts by writing that down before deployment, not after.
What data is this trained on, and who does it represent?
Rural, Tribal, and community populations are frequently underrepresented in the data behind healthcare tools. Leaders serving those populations have to ask the question explicitly.
What happens at the bedside on a bad day?
Evaluating a tool by its demonstration is not evaluation. The relevant test is how it behaves when staffing is short, connectivity is poor, and the workflow is under pressure.
Framework
Technology Evaluation Framework
Governance questions to answer before any AI or digital tool enters a clinical workflow.
1. Purpose and fit
- What clinical or operational problem does this solve, and how is success measured?
- Does it fit the existing workflow, or does it add steps, clicks, or documentation burden?
2. Safety and evidence
- What evidence (validation studies, FDA status, peer benchmarks) supports its accuracy in a population like ours?
- What happens when it is wrong, and who catches the error?
3. Equity and bias
- Was it trained and tested on rural, tribal, older, and minority populations?
- Could it widen care gaps for any patient group?
4. Accountability
- Who owns the tool clinically, operationally, and technically?
- Does a licensed clinician retain final decision authority, and is that documented?
5. Privacy and security
- Is it HIPAA-compliant, with a signed BAA?
- Where is data stored, who can access it, and is patient data used to train the vendor's models?
6. Regulatory and compliance
- Does it meet CMS, state survey, and payer documentation standards?
- Is its output auditable and defensible in a survey or legal review?
7. Workforce readiness
- What training, competency validation, and downtime procedures are required?
- Did frontline staff help evaluate it?
8. Monitoring and exit
- Which KPIs will be tracked after go-live, and how often are they reviewed?
- What are the criteria and process for pausing or removing the tool?
Decision ruleIf any answer in sections 2, 4, or 5 is unknown, the tool does not go live.