Why it matters
Health systems are applying AI to operational problems such as scheduling, patient flow and triage, where improvements free clinical time without making diagnoses.
Operational AI can help match capacity to demand, for example by predicting discharges or missed appointments. Its success depends on integration with clinical workflows and on staff trust in the predictions.
By the numbers
81%of US physicians used AI in their practice in 2026, up from 38% in 2023 1
1,692physicians surveyed by the AMA in 2026 1
2023
2024
2026
Source: American Medical Association, Physician survey on augmented intelligence, 2026
What's changing
- Predictive models estimate demand, admissions and length of stay.
- Digital triage tools route patients to the appropriate care setting.
- Ambient documentation tools reduce administrative work for clinicians.
How leaders are responding
- Starting with operational use cases that have clear measures of success.
- Involving clinicians in design and evaluation.
- Monitoring models for bias and drift across patient groups.
What good looks like
- AI used first on capacity problems with clear measures, such as scheduling
- Clinicians involved in design, testing and sign-off
- A clinical safety case for every AI tool in use
What to watch
- Ambient documentation in clinics and hospitals.
- Evidence requirements for AI tools in health services.
- Integration with electronic health records.
Where to start
Choose a bottleneck with clear data, such as outpatient scheduling, and test an AI tool against current practice with clinicians involved.
Questions for your leadership team
- Where does capacity pressure cost us most?
- How will clinicians come to trust and act on predictions?
- How do we monitor fairness and performance over time?
