💡HTN | Community Brain Trust | 7/18
🧵TOP THREADS OF THE WEEK
In case you missed them, here are highlights of a few interesting conversations from different channels:
Threads included below:
- Claims adjudication timeline
- Medical directory compensation benchmarks
- Best practices for Series A valuation analysis
- Best LLM APIs for healthcare use cases
- Gamification in healthcare
1. Claims adjudication timeline
Q: Feeling silly not knowing this but how long does claims adjudication usually take after submission? Two weeks?
– Brendan Keeler | via #buildersask
Thread Summary: The crew unpacks the actuarial analysis behind claims adjudication and highlights several disruptions that can oftentimes delay the process.
Aaron Neiderhiser: Actuaries do an analysis called IBNR (incurred but not reported) which involves an analysis of services that have been rendered but not yet billed/adjudicated. It’s been a while but I remember something like 95% of claims being paid after 3 months. There is a long tail effect for sure. I would guess the average is something like 1-2 weeks like Tamra says. Inpatient generally longer and varies by payer.
2. Medical director compensation benchmarks
Q: Can anyone share information about what a full-time medical director's compensation looks like for a startup? We're a seed-stage company trying to benchmark the team but we're running into some issues with the clinical side.
– Amanda Stavinga | via #buildersask
Thread Summary: HTNers provide benchmark data for the CMO, Founder role at various company stages - including $0 - $5M; $20 - $100M raised, and more. The group also considers the value of hiring a fractional medical director as an early-stage alternative.
Chris Saxman: Actually, I have that data. See attached. [Additional data for various company stages included in Slack thread here]
3. Best practices for Series A valuation analysis
Q: We are a seed stage startup preparing to raise a series A round. How much time should we spend preparing our own valuation analysis in advance?
On one hand, at this stage the range of potential valuations seems small: we have less than 10 paying customers, some revenue but not millions, and a functioning product. So not worth spending too much time on our own analysis.
On the other hand, I assume valuation is something a VC would negotiate. It seems naïve not to have our own analysis to fall back on.
Anyone have experience here?
– Anonymous Bot | via #buildersask
Thread Summary: The nerds offer up perspectives and benchmarks for what Series A valuations typically look like, as well as how founders can prepare a sound analysis going into the fund raising process.
Michael Lesh: At your stage, valuations are determined by the investors, more so than by your own calculation. Look at comps (Crunchbase, Pitchbook) for companies in your space and stage (but note that valuations are about half what they were a yr ago) for what you might expect. Ideally, you’d run a process and get multiple TSs to be able to negotiate valuation.
In some SaaS models, you can figure valuation is somewhere between 5-10 ARR, but if you don't have clear PMF that may not be an accurate metric. No investor will take risk adjusted DCF very seriously so wouldn’t spend time on that.
Also: many entrepreneurs concentrate too much on top-line valuation, whereas structuring, control provisions, and option pool size need to be considered these days. Plain vanilla is not seen much anymore. And the value of quality investors who share your vision and culture cannot be overemphasized.
4. Best LLM APIs for healthcare use cases
Q: Question for those building product with LLMs. Which LLM APIs (OpenAI, Anthropic, open source options) and vector databases are folks using for use cases where PHI needs to be shared with the vendor? Thoughts on pros/cons of each option? In general, would love to understand folks tech stacks for their LLM applications and how they decided on that stack.
– Anonymous Bot | via #topic-ai-ml
Thread Summary: HTNers discuss advantages and disadvantages to current LLM models on the market - considering APIs from the likes of OpenAI, Pinecone, AWS and more.
Morgan Jeffries: We’re looking at OpenAI models through MS Cognitive Services, largely because we have an existing relationship and BAA. OpenAI offers their own service (also hosted on Azure) and will supposedly sign BAAs, although I’ve read about people having trouble reaching their sales team to discuss it. Anthropic will sign BAAs and has a trust portal where you can get more info. Google claims their BAAs cover Vertex AI, but the Mayo is the only place I know is using them.
5. Gamification in healthcare
Q: What are your favorite healthcare gamification initiatives / companies with compelling gamification strategies?
– Noorjit Sidhu | via #random
Thread Summary: The crew lists out a handful of companies working on gamification strategies within the healthcare space - notable mentions include Well, Irrational Labs, and CareCognitics.
Jake Butler: I go back to Dan Ariely and particularly his work with irrational labs when I'm looking for this kind of inspiration [Irrational Labs link]
Another one came to mind - MadPow Design. They started the Healthcare Experience Design conference out in Boston and you can probably find some good stuff in their archives
Here we highlight a question from the Slack that needs some additional community insights - if you have a helpful thought, jump in below!
Q: Does anyone know of any good clinician utilization tracking tools / methodologies for both full time staff (w2) as well as independent contractor models? Thank you.
– Brian Kinsella | via #builderask
🤖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!
Here we highlight helpful resources from across the community:
- The Paths to Healthcare PLG (Part 1) by Ben Lee – Part 1 of a two part series unpacking why product-led growth is hard to come by in healthcare and what, if anything can be done about it.
- Large language models encode clinical knowledge via Evan Brociner – Google released a paper in Nature reviewing its state-of-the-art LLM for medicine, Med-PaLM.
- Tackling healthcare's biggest burdens with generative AI via James Lu – A summary from McKinsey on various healthcare use cases for gen AI.