Hey friends! 👋 Excited to share a demo of an AI-powered telehealth feature we built on top of Medplum! In this demo, you’ll see a patient book an appointment via the patient portal, followed by a video consultation. AI automatically generates a draft Clinical Note and Prescription, which the doctor can quickly review and finalize. The prescription is then delivered instantly to the patient. 🎥 Watch the demo here: It’s amazing how Medplum has simplified the healthcare app development! Would love to hear your thoughts and feedback.

Very cool Santosh! How long did this implementation take roughly? (Dev time I mean)

Thank you, Reshma! The telehealth feature took about a month to develop, preceded by a couple of weeks of conceptualization and UX design. Using Medplum's FHIR dev platform really made it so much easier, allowing us to focus more on the frontend with minimal backend effort.

TLDR version: 👏 ! @Santosh Chandak beautiful showcasing of medplum, and your work continues to inspire me personally. Years ago (before EHRs!) I imagined someday patient/clinician encounters would be fully recorded and transcribed to provide accountability for both patient and clinician. Medical charting continues to suffer from the 'clinician bias' and 'patient fear'; what was important in the encounter was not documented, or what was not important is over emphasized in the clinician's final note. Your showcase underpins an implicit acceptance of accountability, which is a massive step forward. Interestingly if I bring a digital recorder into an examination room, trust and implicit confidentiality vanishes. I've never been an advocate for telehealth, but here we are today and telehealth is not going away, ever. The mock encounter in your showcase does an excellent job showcasing your work. I could not help but fear for the mock encounter outcome however. Critical thinking and audit control of the (potentially deadly outcome) mock encounter is pervasively absent; this mock patient could very well suffer from a deadly MI during his sleep (HTN+substernal chest pain and high risk for Sleep Apnea), and this places your mock encounter as an easy case for medmal. I hope to see your work on github because you are moving the needle; but with that great talent of you and your team comes great responsibility too. I look forward to more of your brilliant showcasing of medplum!

Hi Kai, Thanks so much for the kind words! In terms of effort, we had a couple of developers working on and off over the course of a month—roughly 1.5 FTE for the telehealth module. Some foundational pieces, like the AI-generated prescription, were already developed as separate modules. As for estimates, I have to say—using Medplum along with AI-assisted coding has significantly accelerated our development process!

@Santosh Chandak - do you use Cursor, Claude code or something else?

Hi Reshma, We’re mainly using Cursor. As AI understands the standard FHIR data model very well, one can produce great results working with Medplum in quick time.

I really liked the demo, Santosh 😃 It's a very good feature for telemedicine. At a hospital where I work, they created a feature very similar to this one, with the difference that the audio is recorded on a mobile device during a face-to-face consultation. Many doctors refused to use it because patients feel nervous about being recorded so obviously. I think it can be applied much better in telemedicine, as the demo shows.