💡 HTN | Community Brain Trust | 8/22
🧵TOP THREADS OF THE WEEK
In case you missed them, here are highlights of a few interesting conversations from different channels:
Threads included below:
- Understanding EHR interoperability exceptions
- How-to: set up an SFTP server
- Synthetic medical records for LLM experimentation
- Transitioning from a fintech to health tech product role
- Integrated devices market overview
1. Understanding EHR interoperability exceptions
Q: Yo! Looking for an interoperability expert. We have a provider facing app and need to validate that the provider is working with a patient such that we can pull that patient's files via particle. Is the best way to do this by having the provider authenticate through their EHR? Have y'all tried anything else when building with particle? Thank you!!!
– Adam Steinle | via #buildersask
Thread Summary: The HTN crew jumps in with advice on how to handle provider-patient validation under a new exception pathway under the Carequality framework.
Dave Boerner: You should probably attempt to build the business logic in to the workflow to ensure Hona is just pulling the data for the properly attributed patients to the requesting provider. Particle doesn’t actually even require you to send the NPI, it’s optional with them but probably best to include anyways.
As BK mentioned you’ll need to also reciprocate with some new data so something net new should have happened, like the clinician had an encounter with the patient and consulted Hona patient summarization to get up to speed.
2. How-to: set up an SFTP server
Q: SFTP setup tips?
Hi! We’re trying to set up a process to receive eligibility files from a payer via SFTP. I’ve never set this up before so I’d love some tips. In particular:
1. Any recommended providers that can host an SFTP server for us so that we don’t have to maintain it ourselves? Ideally one of the standard cloud providers (Google, etc.)?
2. How do you ensure that your SFTP server is up to date, secure etc., unless it is maintained by some vendor?
3. Anything else I should know here?
I speak engineer so feel free to use technical terms if you want.
– Sameer Madan | via #buildersask
Thread Summary: The techie Nerds jump in to share a few data resources, plus advice on how to approach the set up of receiving eligibility files via SFTP.
Michael Herrick: Yuck! Fortunately, it’s pretty easy to set up and secure.
1. A virtual machine on Google Cloud Platform would be easy to set up. Schedule it to turn itself off after hours. If you know the other side will be transmitting at a specific time, run it only for that period of time. The cost will then be minimal.
2. Also easy to host it yourself. Do it in a virtualized environment with no other processes running. That way, you can turn on automatic OS updates and security patches with little fear that they will cause some other mission-critical process to break.
3. Authentication by a long, random, unmemorized text token (it should not be called “password” when it is not memorized or entered by a human). A random SHA hash would be fine. Make sure you share the key via a super secure transfer medium, and delete it from the transfer medium afterwards. Don’t email it.
4. No need to worry about expiring the key. If you make it long enough, it will be highly secure against brute force attacks. Of course, do change it if there’s ever evidence of compromise. If you can automate the expiration and replacement, that’s nice but not a must-have.
5. Multifactor is not relevant because it’s not a password and you don’t have access to other factors (human knowledge, human biometrics) in an automated connection.
6. Add network level protection. Allow access only from a pre-registered IP address.
7. Throttle connection attempts. There should almost never be an unsuccessful connection attempt, especially after initial setup, so any time you see 3 failed logins, shut it down and alert an admin.
8. Log all access.
3. Synthetic medical records for LLM experimentation
Q: Similar question! Anyone know where I can get de-identified or synthetic raw transcripts of doctor-patient encounters? I want to play around with using GPT4 or Anthropic for converting them into formal notes.
– Gabe Strauss | via #topic-ai-ml
Thread Summary: A discussion offering up multiple synthetic medical data resources, as well as brainstorming around how to generate sample medical records using ChatGPT.
Morgan Jeffries: @Gabe Strauss maybe I’m way off, but I think you’d need to synthesize that. It wouldn’t be part of the EHR, and I don’t think the companies that have those transcripts have an incentive to release them. Curious to hear if anyone has any ideas, though.
Gabe, one trick here is that people don’t always speak in complete or grammatical sentences, and they may use filler words. I got more realistic dialogue when I called that out in the prompt.
[Example:] “Can you create a short transcript of the first ~30 seconds of a physician-patient encounter where the patient wants to discuss a vesicular rash along their right side? It should be a realistic transcript of actual human dialog. They may or may not speak in complete and/or grammatical sentences, and they may use filler words (e.g., “um,” “uh”) or stammer. The patient is college-educated but has no medical training and will not use specialized medical vocabulary. The encounter starts when the physician enters the exam room where the patient is sitting. The dialog only needs to include their brief introductions and everything up to the patient’s initial description of their problem.”
4. Transitioning from a fintech to health tech product role
Q: Hi HTN Product people,
I'm making a domain switch to healthtech from fintech. I'm curious:
What do you wish you knew BEFORE working in healthtech?
Thanks for sharing your perspectives!
– Sara Edwards | via #topic-product
Thread Summary: An interesting thread sharing experiences of product folks moving from other industries into health tech - thing they wish they knew going in, advice, and broader observations of the role of product in health tech.
Seb Agertoft: I made the same switch Sara. Lots of similarities between the two - heavily regulated industries, big old-school incumbents, high-stakes products.
In my opinion, healthtech is still very much playing catchup with fintech with regards to funding, density of high quality product companies, and as a result the density of talent (this community excluded).
Depending on what company you're going into, it may be that the product is just one part of an overall healthcare offering alongside clinicians for example. If that's the case I'd say spend a lot of time understanding the overall service design
5. Integrated devices market overview
Q: My friend has a question about integrating devices (think blood pressure cuffs, pulse ox, weight scales, etc etc) into a platform or an EMR. A while back Validic was the clear leader in this area but it's been a while. Anyone has any ideas?
– Matthew Holt | via #buildersask
Thread Summary: HTNers discuss companies building in the integrated devices space - calling out top companies, reviewing vendor offerings, and watchouts for developers to be aware of during integration.
Brendan Keeler: The Plaid for wearables market:
- Human API
The above all are a small widget to authorize the pull of data and then the API to actually pull it. It's good for startups / apps to build against.
Validic started with that, but now notably has a SaaS application for hospitals, as hospitals don't typically have the appetite to use the API.
Pushing vitals data into an EHR is a different ballgame. You'd be looking at traditional EHR integration companies (Redox, Healthjump, etc).
Here we highlight a question from the Slack that needs some additional community insights - if you have a helpful thought, jump in below!
Q: Is anyone familiar with how clinical trial software products are priced?
For example, how do companies price site recruitment and activation applications/services?
– Haroon Zahid | via #buildersask
🤖HTN KNOWLEDGE BOT
If you have your own question(s) to ask, don’t forget that a good place to start is our HTN Knowledge Bot. It’s our smart search tool that makes it easier to access the wisdom shared within the HTN powered by ChatGPT. You can log in and use it on the website (here) or see how to use it directly in Slack here.
Check out the example ask below!
I want to learn more about the Medicare care at home space
Knowledge Bot Answer:
I found some relevant information for you on the Medicare care at home space:
1. According to a conversation in the Health Tech Nerds Slack channel, Medicare Advantage plans have been slowly adopting the opportunity to pay for home environment services as a supplemental benefit, including home modifications for “fall proofing” a member’s residence. However, the penetration of these services is still relatively low at around 10%, but it is increasing. You can read the conversation here.
2. There is an article about two providers joining forces to launch a virtual in-home care company. This article provides insights into the use of technology in delivering care at home. You can read the article here.
3. In another conversation in the Health Tech Nerds Slack channel, someone asked if anyone is in the Medicaid Home Care Space. One response mentioned that the Medarrive folks are in that space. You can read the conversation here.
4. There is also a conversation in the Health Tech Nerds Slack channel where someone asked for good overviews of the home health industry. While there was no direct response to that question, it indicates that there is interest in learning more about the home health industry. You can read the conversation here.
These resources should provide you with a good starting point to learn more about the Medicare care at home space.
Here we highlight helpful resources from across the community:
- 2024 Proposed MPFS - Care Management Reimbursement by Kaitlyn O'Connor - An in depth calculation of CMS' proposed reimbursement rates for RPM and RTM for 2024.
- Handwave.ai via Evan Brociner - Handwave.ai is an AI fundraising network, comprised of 20 top AI investors. Founders can submit their pitch decks to the site to have it reviewed by the cohort of investors all at once.
- The Impact of AI Documentation on Reducing Clinician Mental Burden and Increasing Time Available for Care Interactions. by Andrew Lundquist - This article examines the role of AI documentation in the healthcare sector and its potential to alleviate the mental burden on clinicians while improving efficiency and saving time.