Face-to-Face (F2F) AI

Dr Ben Brown
Dr Ben Brown
  • Updated

What is F2F AI? 

F2F AI suggests which patients may need a face-to-face or in-person consultation.

For example, these might be patients who need a physical examination. F2F AI does this by displaying a 'people' icon against the request in the inbox, and by pre-populating the 'actions' triage decision for you with ‘See patient face-to-face’.

This triage option is only visible to clinical users, because whether a patient needs a face-to-face consultation should be a clinical decision, and we use these decisions to teach and monitor Patchs AI.

F2F AI's suggestions are meant to assist, not replace clinical judgment. You should therefore still triage and reply to all PATCHS requests ASAP.

Inbox view

Triage decision view

What are the benefits of F2F AI?

The idea of F2F AI is that you may wish to book these patients in directly for a face-to-face appointment, rather than dealing with them remotely first via Patchs or by telephone, and then making the decision to bring them in on top of that. This could:

  • Avoid workload duplication by reducing the number of contacts you have with the patient for the same request
  • Improve patient safety by seeing patients that need an in-person appointment quicker 
  • Help prioritise your workload by highlighting which patients need an in-person consultation 
  • Reduce your workload by helping you make decisions about whether patients need an in-person appointment more easily

How accurate is F2F AI?

The most important thing for F2F AI is its ability to identify patients that need an in-person appointment. This is because patients it suggests need an in-person appointment but don't (false positives) can be dealt with in-person, but patients it suggests don't need an in-person appointment but do (false negatives) can't be dealt with remotely.

The best way to measure this is the percentage of request that need an in-person appointment it gets right, which is called the sensitivity. The current version of F2F AI has a sensitivity of 92%. Nothing is 100% accurate, even human GPs. It is not known how well human GPs perform the same task as F2F AI so we can compare them, but we are conducting research to find out.

Further detail on the performance of all AIs is given in the Clinical Safety Case Report.

What should I do if I see a suggestion from F2F AI?

If you agree with the suggestion made by F2F AI then you should leave the 'See patient face-to-face' option checked or unchecked.

If F2F AI thinks a patient needs a face-to-face consultation when they don’t (false positive)… You should untick the 'See patient face-to-face' option when making your triage decision.

If F2F AI misses a patient that needs a face-to-face consultation (false negative)… You should tick the 'See patient face-to-face' option when making your triage decision.

Because Patchs is constantly learning, this will help it correct its mistakes in future.

As mentioned above, the ‘See patient face-to-face’ option is only visible to clinical users. This is because whether a patient needs a face-to-face consultation should ideally be a clinical decision, and we use these decisions to teach and monitor Patchs AI.

Was this article helpful?