← Back to work

50% increase in clinical capacity by reducing admin time.

Outcomes.

  • 50% increase in clinical capacity unlocked with the same number of clinicians.

  • 89% adoption rate for the report writing assistant.

  • Up to 75% reduction in report writing time across 18 service lines.

  • Most multi-disciplinary team meetings (MDTs) are now completed asynchronously, reducing service delivery time. Calendar conflicts were previously delaying MDTs by 2–3 weeks on average.

  • ~30 minute reduction in meeting prep time per MDT.

  • All MDT communication kept as a secure, auditable record within the clinical platform.

Problem.

Healios wanted to increase clinical sessions per clinician per week from 4 to 6, but were unable to do this due to inefficient clinical admin tasks slowing things down.

What I did.

AI report writing assistant.

AI-assisted report section summaries transformed post-assessment report writing — one of the most time-intensive parts of a clinician's day. The tool structures the information-gathering process, generates clinical write-ups from clinician-controlled inputs, and produces consistent output with correct spelling and grammar baked in.

It's optional, and can be managed per service if a trust or partner doesn't want AI used for a particular service. Service-specific prompts keep the output consistent, and paper notes can be added directly.

I worked with our chief clinical officer, clinical safety officer and head of information security to make sure this tool was compliant and safe to use.

Asynchronous multi-disciplinary team meetings.

I worked with our clinical head of assessments to identify ways we could make the MDT process asynchronous by default, removing the calendar conflict problem and the need to do meeting prep.

In most cases clinicians agree on the majority of DSM-5 criteria and notes. Email or Slack is enough to resolve the remaining issues. MDT meetings are still held for complex cases. I designed this new flow end to end and worked closely with our engineers and a technical clinician to implement it for final testing.

Learnings.

  • AI regulation in clinical settings is complicated and still evolving. It also overlaps with guidance on clinical devices, making it more complex and potentially costly to get wrong — if I worked in this area again I'd have a head start, but there is much more to learn.

  • Changing two core existing clinical processes required more collaboration, planning and change management outside of the product than I expected — I'd be more prepared for this in future.

  • Clinicians with a background in the NHS are not as used to or comfortable with change as the average user — I would allow even more time in future for adoption and discussion on risk and safety concerns to allay fears.