EHR Auto Documentation is Great, but Augmenting Doctors Is the Future

If you read Healthcare IT Today regularly, or listen to our health it podcast, then you’re probably tired of hearing me talk about the potential of Ambient Clinical Voice in healthcare.  For good or bad, I’m probably not going to stop.  At least not until it’s proven that it can’t ever fulfill it’s potential.  However, there are already promising signs that this will never be the case.

I don’t know any clinician who wouldn’t love for the EHR to auto document the visit for them.  That’s the promise of ambient clinical voice and related technologies.  There’s no doubt some trepidation about whether it really works which is healthy, but once it’s ready for prime time it’s hard to imagine every doctor not wanting it.  Can you imagine working with ambient clinical voice at one healthcare organization and then moving to one that didn’t have it?  That would be shocking to the doctor.

While ambient clinical voice is great for being able to auto document the note and relieve clinicians of that documentation burden, I’m seeing that this is just the first step.  The next step past automating the documentation is a casual AI listener that’s informing, prompting, and possibly even teaching the doctor.

How reassuring would it be for a doctor to be able to look down at their phone or on their computer and see that the AI engine is confirming the diagnosis they have in mind based solely on their conversation with the patient.  Plus, it could suggest other possible diagnoses to consider.  At least for now, it’s not going to make the decision for the doctor, but just the prompt is going to force the doctor to really verify their diagnosis.

Of course, this gets really tricky really quickly.  If the AI sends poor prompts, then the doctor will just ignore it.  So, the prompts and information need to be relevant.  However, relevant prompts that don’t cause alert fatigue could even be welcomed by a doctor who is working quickly, confusing patients, burnt out, etc and may have missed something.

Plus, this type of technology could be powerfully used when training new doctors.  It will cause the age old learning question of what do we need to learn and what can we rely on Google (or in this case AI) to know, but that’s a classic evolution of learning that’s happened forever.  I’ve heard many doctors share how clinical decision support educated them on an alternative diagnosis or course of treatment to consider when handling a challenging case (which is every visit for a new doctor).  This is just doing the same thing in a more automated way at the point of care.

Where this evolution gets even more interesting is when this casual AI listener spends time listening to the patient before the visit and before the doctor even sees the patient.  Now, the patient talking to the AI listener could be doing a bunch of the documentation for the doctor.  Plus, the doctor could have the AI listener present it a summary of what the patient has shared in the format they prefer.

Certainly this approach won’t replace the patient-doctor interaction.  The doctor will still need to hear and observe the patient.  However, getting the mundane interaction completed by a casual AI listener could actually enhance the time the doctor spends with the patient.  The doctor can verify key findings and then spend more of the time interacting with the patient on what matters most.

Taking this one step further, eventually this casual AI listener is going to take place at the new digital clinical front door.  Will we get to the point that the casual AI listener will get so good that it only sends patients to the doctor that really need the doctor’s help?  Plus, when the patient arrives at the doctor, the casual AI listener will have completed a deep dive into the patient’s health that it can present to the doctor upon arrival.

I get that this is all a bit hard to process since today’s healthcare experience feels so dramatically different than what’s described above.  And it will take time to fully realize everything that I’m laying out here.  However, for those familiar with ambient clinical voice, this isn’t some idea out of a sci-fi movie.  It’s the natural evolution of this technology.

In fact, the thing holding this back is probably more cultural than it is technical at this point.  However, cultures can change quickly with the right motivation.  Automating EHR documentation is where it’s going to start since it solves a problem everyone can see and feel.  However, don’t be surprised if it quickly evolves to augmenting doctors in really exciting ways.