Health
The slowest part of a medical consultation
4 min read
We've walked into a lot of private clinics, and we keep finding the same thing: a patient's history scattered across PDFs, scanned letters, and free-text notes. A blood panel from two years ago. A referral letter. A consultation note where a colleague jotted down a blood pressure reading, one that would be useful to track over time, if anyone could find it again.
The information exists. It's just unreachable. And that gap is quietly expensive.
What "unreachable" actually looks like
We've spent the past year talking to doctors in the UK, and the same three frustrations come up again and again:
The PDF problem. Blood results and diagnostic reports arrive as PDFs. You can open them, but you can't search across them; and you can't see how a single marker has moved over the years without opening every file and reading each one by hand.
The note problem. Consultation notes are written, saved, and then effectively lost. Last spring's blood pressure reading is in there somewhere. So is the medication change, the family history, the passing observation that turned out to matter. None of it is retrievable in any practical sense.
The tooling problem. Even when a clinic adopts an AI scribe, it usually produces yet another block of text the existing EHR can't do anything with. You've added a tool, not solved the problem.
All of this slows down the everyday work of a consultation. Even calculating a familiar risk score like a QRISK, starts with pulling several values out of different documents by hand, so a routine task begins with a search rather than an answer.
A compounding problem
At first, this looks like a manageable nuisance. But it compounds. As more patients and more data pile up in the clinic's systems, what was an annoyance becomes unmanageable.
The common instinct is to throw the files at ChatGPT or another general-purpose LLM and let it find the answer. There are two problems with that:
Cost. Running a fresh query every time you need an answer adds up, in both time and money. Big tech has subsidised these queries to keep them looking cheap, but more providers are now adding usage limits and raising prices.
Reliability. General-purpose LLMs are stochastic by design, and the more raw, messy data you feed them, the more room there is for error. That's not a foundation you want under a clinical workflow.
The real issue is the structure of healthcare data itself, and it isn't unique to any one clinic. Roughly 80% of healthcare data is unstructured: it lives in physician notes, clinical narratives, and imaging reports rather than in tidy database fields. Most electronic health records were never built to handle that kind of data, so it often ends up ignored or abandoned. (Applied Clinical Trials Online)
The cost shows up as time. In one study of primary care, physicians spent an average of 36 minutes in the EHR for visits scheduled to last 30, much of it chart review, hunting for information that was already on file. The American Medical Association has found that physicians spend nearly twice as much time on EHR and desk work as they do with patients. (American Medical Association)
For a private clinic, the arithmetic is direct: time spent digging through old files is time not spent with patients, and not spent at home, either.
What we do instead
Nudge Care takes the documents you already have, the PDFs, the scanned diagnostics, the free-text notes, and runs them through pipelines that turn them into structured, organised entries. A lab value becomes a data point you can chart over time. A diagnosis becomes something you can search and filter. A note becomes queryable.
One detail matters most here: nothing is taken on trust. Every structured entry can be reviewed and verified by a clinician before it's used. The pipeline does the extraction; the clinician keeps the final say. You get the speed of automation without giving up the accuracy your work depends on.
And once the data is structured, it stays yours, inside your own practice, in your own workflow. Established risk scores can be calculated from values already on file instead of being typed in by hand, and reports can be generated straight from structured fields. What you do with that information is a clinical judgement; Nudge's job is to make sure the inputs are complete, verified, and in front of you.
Who this is for
We work with two kinds of clients:
Private doctors and clinics who want a patient's history to be searchable and visual instead of buried, so a consultation starts with the full picture already on screen, not a stack of files to open.
Companies building clinical software who need reliable, structured data underneath their product and can't absorb the accuracy gap that comes from working with raw, messy text.
In both cases the principle is the same: everything needed for a well-prepared consultation is already sitting in your files. The job is to make it reachable, accurate, and verifiable, without adding work to anyone's day.
Want to see what your clinic's data looks like once it's structured? Book a demo.
Sign up today
