Industry
AI in healthcare
AI in healthcare administration and operations: documentation support, scheduling, coding, prior authorisation and patient communication — with clinical safety and GDPR constraints.
Miguel Torres, Founder, Merjora · Updated 3 September 2026
In short
The clearest AI value in healthcare is administrative: clinical documentation support, appointment and referral handling, coding assistance, prior-authorisation preparation and patient communication. These reduce the non-clinical burden on staff without entering the regulated territory of diagnosis or treatment recommendation, which carries medical device and EU AI Act obligations that change the project entirely.
Context
Clinicians and administrators spend a large share of their time on documentation, coordination and correspondence. That burden is where AI can help now, with a clear human decision point retained.
Anything that influences diagnosis, triage severity or treatment moves into regulated territory. That does not make it impossible — it makes it a different project with different obligations, evidence requirements and timelines.
Where the opportunities are
Clinical documentation support
Drafting notes and letters from consultation audio or structured inputs, with the clinician reviewing and signing. The clinician remains the author of record.
Complexity: Moderate
Appointment and referral handling
Structuring referrals, handling scheduling correspondence, and sending reminders and preparation instructions.
Complexity: Low
Coding and billing support
Suggesting codes from documented encounters for coder review, with rationale and source text shown.
Complexity: Moderate
Prior authorisation and payer correspondence
Assembling the evidence pack and drafting submissions, a large administrative burden in payer-based systems.
Complexity: Moderate
Patient communication
Answering logistical and administrative questions, with clinical questions routed to staff by design.
Complexity: Low
Constraints that shape delivery
- Special-category health data under GDPR: lawful basis, minimisation, retention and processor location all need explicit positions
- Clinical safety governance applies to anything touching care pathways
- Medical device and EU AI Act obligations attach to diagnostic or triage-influencing functionality
- Clinician trust is earned through reviewable output and easy correction, not accuracy claims
- Integration with the electronic patient record is usually the largest cost line
ROI considerations
- Documentation time per encounter × encounters is the primary driver; clinician time is the highest-value hour in the organisation
- Administrative rework and payer rejections are a measurable secondary driver
- Include clinical safety review and information governance in first-year cost
- Capacity value — more appointments per session — should only be claimed if the schedule actually changes
Documented examples
Publicly documented industry examples. These companies are not Merjora clients.
- Acentra Health →
Publicly documented AI use in healthcare administration.
Frequently asked questions
- What are the safest AI use cases in healthcare?
- Administrative ones: documentation drafting with clinician sign-off, scheduling and referral handling, coding support and logistical patient communication. They avoid clinical decision-making while removing real workload.
- Does AI in healthcare require regulatory approval?
- It depends on function. Administrative support generally does not, while anything influencing diagnosis or treatment can fall under medical device rules and higher-risk EU AI Act obligations. Classify the intended purpose before building.
Where would AI pay off in your healthcare business?
Merjora maps your workflows, scores each opportunity and quantifies the value range — with the constraints named.
Discover your AI opportunitiesRelated reading
Editorial standard. Merjora publishes analysis, frameworks and publicly documented examples. We do not publish invented statistics, unattributed benchmarks or unverified customer stories.