Adaptive Learning is an experimental feature, managed from Settings → Heidi Labs.
Adaptive Learning is how Heidi learns from the way you edit your notes, so you stop making the same correction twice.
Every clinician documents differently. You might shorten "hypertension" to "HTN", or always correct a mis-spelled name to the clinician you actually work with. Adaptive Learning notices the corrections you make repeatedly, learns them, and applies them to your future notes before you start editing.
Adaptive Learning is in Heidi Labs (beta). As always, review every note carefully before you rely on it.
How does it work?
At the end of the day, Heidi reviews the edits you've made to your notes and looks for corrections worth learning. It waits until it's confident a correction reflects how you want to document, rather than acting on a one-off edit.
Each learning keeps the context it came from. It records when it should apply, when it shouldn't, and, where your edits show a real boundary, which templates it belongs to. If you write "Pt" in your quick SOAP template but "patient" in your detailed consult template, Heidi learns exactly that. Nothing is applied as a blind global find-and-replace.
When you generate a note, Heidi checks it against your learnings and applies the ones that fit. You can see exactly what changed in the note's version history.
Heidi unlearns, too. If it applies a change you don't want, edit it back. The next learning run treats that as a signal and either retires the learning or narrows where it applies. Your edits are always the source of truth.
When will Heidi pick up my corrections?
Learning runs once a day, overnight. A correction you start making today will usually show up in your notes from tomorrow. Allow up to 24 hours for a new learning to take effect.
How do I turn it on?
Go to Settings → Heidi Labs and switch on Adaptive Learning. Heidi reviews your recent edits that night, and starts applying what it learns to your notes from the next day.
Because it's an experimental feature, you'll be asked to read and accept the Experimental Features Supplemental Terms of Use the first time you turn one on. If you already had Adaptive Learning on, you'll be asked once the next time you open Heidi. Declining turns Adaptive Learning off. You can turn it back on whenever you like, and you'll be asked to accept the terms then.
A few things to know
Notes may take slightly longer to generate.
Heidi reviews each note against your learnings before showing it to you, and that adds a little time. We think a note that needs fewer edits is worth the wait, but if you disagree you can toggle the feature off in Heidi Labs.
Adaptive Learning replaces Personalisation for your notes.
It is the next generation of that feature: learnings are precise corrections rather than broad style hints, they respect the context and templates they were learned in, and they retire themselves when you edit a change back. While it's on, your notes are personalised by Adaptive Learning alone, and your previous Personalisation preferences no longer apply to your notes, though they continue to apply to your documents. You won't get two systems making overlapping changes.
You stay in control.
Every applied change is recorded in version history, and you can switch Adaptive Learning off at any time in Settings → Heidi Labs.
Privacy
Adaptive Learning learns your documentation preferences, which means the way you edit. Your learnings are private to you: Heidi builds them only from your own edits and applies them only to your own notes.
Heidi learns from your notes while Adaptive Learning is switched on. If you turn it off, Heidi stops applying corrections straight away.
Adaptive Learning operates within Heidi's existing data privacy framework, and the Experimental Features Supplemental Terms of Use set out what applies additionally to experimental features.
FAQ
Why hasn't Heidi learned my correction yet?
Learning runs daily, so allow up to 24 hours. Heidi also waits until it's confident a correction reflects how you really want to document, so it learns your habits rather than acting on one-off edits.
Will a learning from one template leak into another?
Learnings apply in the context they were learned in. If your edits show that a preference belongs to one template, or a few, Heidi scopes the learning to those templates. A preference you apply everywhere stays available across all your notes.
Heidi applied something I didn't want. What do I do?
Edit it back. Heidi picks the reversion up in the next daily run and retires or narrows the learning. If you want it gone immediately, you can turn Adaptive Learning off in Heidi Labs.
Where can I see what Heidi has learned?
There isn't a view of your learnings in Heidi yet; version history shows each change as it's applied. If you'd like to see and manage your learnings, let us know on our feature request board. It helps us prioritise.
What happened to my old personalisations?
You may notice the Personalisation button is gone from your notes. Adaptive Learning has taken over personalising your notes, so your old personalisation preferences no longer apply there.
Does it learn how I structure my notes?
Right now Adaptive Learning focuses on the wording corrections you make, like abbreviations and fixes. Learning structural preferences, such as how you order or format sections, is something we're exploring. If that would be useful to you, let us know on our feature request board.