Launch HN: InpharmD (YC W21) – curated drug information for doctors
My name is Ashish, and I’m the CEO/ co founder of InpharmD (https://inpharmd.com). We take questions from doctors and return curated, evidence - based answers.
I was a clinical pharmacist offering a remote service from a University for 10 years. Ask us anything, we begged, and our team of pharmacists, residents, and students would look it up, get through the paywalls, and provide the answer.
I passed out business cards around local hospitals. They were lost over time.
Then I passed out business cards with magnets. They stuck around, but there aren’t that many places in the hospital with the magnetic surfaces.
Eventually, people stored our number, but we’d ask so many questions when they called, they couldn’t ask theirs: who are you, where are you calling from, what’s your email, spell it, etc, etc, etc. Often, they’d hang up on us, and I don’t blame them. The average doctor now sees five patients an hour.
I realized I wasn’t alone, and hundreds of other academicians, all leading their own teams, had the same problem. So, we formed a network and interviewed hundreds of our customers about how they’d ideally interact with us. What we needed to build was simple: one touch request.
My co - founder Tulasee built that and since, we learned that AI can transcribe PDFs faster (but not yet better) than our pharmacists. We started with 5,000 of our own study abstracts, assigned weights for corresponding content in their respective PDFs, and now we continuously reassign the weights until the algorithm can completely make our own abstracts. Our latest test revealed 94% accuracy against a matched human control, but with medical information, this will need to be 100% before we can rely on it.
We think Watson was a missed opportunity, so we called our algorithm Sherlock. We’re launching a partnership with the American Society of Health-System Pharmacists® (ASHP- https://ashp.org), using their database of 1,300 vetted drug monographs, so Sherlock can field questions at the point of care.
We’ve been fortunate to find early adopter health systems to pay for our service: WellStar, Ochsner, University of Maryland, Georgia DPH, and St Francis. We’re typically compared to the cost of their healthcare providers manually searching, and we end up cheaper.
We love this community and we’d welcome your ideas/ experiences/ feedback on what we’re building!
73 comments
[ 2.2 ms ] story [ 137 ms ] threadAnd it comes down to liability as well. Basing your treatment decisions on UpToDate or NCCS guidelines or Cochrane Reviews is pretty defensible. But basing it on the findings of an AI start up? How is that being addressed?
But don’t get me wrong. I think there is a lot of value in physicians having easy access to pharmacists to discuss treatment options. Right now that doesn’t happen that easily.
The consequences of wording something wrongly are huge.
To be clear, we stop short of making recommendations. Our goal is to give the provider all the info she needs to make her own evidence informed decision.
Public data + our data shows only ~1/2 of clinical questions can be answered by Up to Date (or references like them).
Those online compendia employ a small team of credible authors + have massive scale, so they focus on the mainstream questions.
Our custom solution is designed around the long tail of emerging or complex questions.
For example, if you want to know about dexamethasone 4mg for early stage COVID, Up to Date will have it. But if you want to know about dexamethasone 20 mg, you’re faced with doing this literature search on PubMed (which = 2 hours + 4 journals) .
We get this question a lot, and really appreciate you surfacing it here. We made this into a FAQ for anyone that prefers to visualize it: https://www.inpharmd.com/faq
Maybe this is how any startup takes on any large incumbent? IDK. We love your collective wisdom, we learn a ton from you all.
For anyone curious , we actually did this question:
https://www.inpharmd.com/is-there-any-data-to-support-higher...
We built our own in house solution to this exact problem, but its maintained by our own clinical staff, it costs us north of $300k a year to maintain it! This only works of course because we have over 150 specialty pharmacies in our 'network' so we have volume to help keep up with costs.
I could honestly see us leveraging a solution like this at some point in the future, and probably anybody running fewer than 10 specialty pharmacies needs a solution like this.
Good luck to you guys!
And WOW re: $300k per year. I’m certain we could do this for a fraction of that cost. Ashish at InpharmD dot com :-)
After reading your description, and watching the video on you website, I can safely say: I have no idea what service you provide.
Adding example question might help.
In Denmark the government office for medicin provide a free service where you’re able to look up all approved drugs, their usage, side effects, treatment plan and so on. It that what you provide or does it go further than that?
An example question is a fantastic idea and will land with both audiences :-)
We’ll get our top three most recent on the homepage ASAP but in the meantime:
If you have a typical late stage COVID patient, a static patient resource like you described is perfectly fine to look at efficacy of the standard dexamethasone 6mg treatment.
But if you have an atypical patient and considering a 20mg dose, you’re out of luck. Most patients are atypical and we see an unmet need with a long tail of atypical questions.
https://www.inpharmd.com/is-there-any-data-to-support-higher...
I would also be careful about how you use the term atypical, in fact most patients with COVID are typical and improve with the standard treatment. :)
But there’s a new school type of doc that realizes how vast the medical literature is (20m studies), how quickly it’s changing (20,000 just added on COVID), and has tons of questions.
All medical references were essentially converted to SAAS products from large books- so they’re thousands of pages covering the mainstream topics, and they answer only ~1/2 of questions.
Health systems employ clinical pharmacists to answer the other 1/2 with evidence because it’s insanely high ROI for them to do so (5M/ pharmacist/ year/ hospital). Health systems outsource this to us because we’re more efficient.
I didn't know this.
In that case, software is probably way more scalable than hiring more and more pharmacists.
We are using AWS's machine learning service called Kendra, which indexes the clinical content, and provide search based on natural language processing. In our testing, it is able to find relevant information with pretty good results. Kendra also has capability for users to rate results so it can become smarter and give better answers over time.
We wrote integration between Kendra and AWS's chat bot service, Lex in a lambda based serverless architecture.
Although with the finely tagged data it may not be necessary.
I asked about the safety of a drug and the product gave me a summary and literature to answer complex clinical questions fast. This product does a lot of the work that before involved me lots of time clicking and searching and trying tons of different Google searches. For a busy physician it works great.
It saved me time and helped ensure I was using the most up to date research for my decision making. This would’ve been a game changer when I was at an academic hospital doing clinical research, but even as a private physician it’s great to ensure you’re always up to date with the current evidence.
e.g. certain antihistamines for sleep aid
Also, hiring?
We just have to be fair and balanced about how we present (which we are anyway) + be extra vigilant to include all relevant prescribing info, and especially the actual indications and any boxed warnings.
As for hiring, yes, and for anyone interested, I firmly believe interest in what we’re building is the most important thing at this point, and everything else is secondary. Ashish at InpharmD dot com
When I was a teenager (and before all answers to everything were online), I went to multiple docs and no one was aware of a particular prescription that, once I found out about it, completely "cured" the issue (Drysol).
Looking back, it was obvious the doctors just weren't aware of a valid treatment for the issue. It'd be great if a service like this helps doctors discover treatments for less than common conditions.
Side note: I recently discovered iontophoresis for the treatment of hyperhydrosis without the use of drugs at all... discovered via a Facebook ad for the machine (don't ask me how Facebook knew to target me for such a specific product). I bought it a couple months ago, and wow - it works. I've been googling the studies around it, which date back to 1950's -- not something new, but also dumbfounded as to how I wasn't aware of this treatment sooner, and why no doctor was aware of it either (or at least didn't bother mentioning it as an option).
Best of luck to you! Lots of problems to be solved in the health space.
I had a fever on a trip in africa where it felt like sweat was pouring out of my skin and it was incredibly uncomfortable
Now, info travels fast, but there’s so much info coming at us, we choke on it.
Totally relate to your story, I come from rural India and saw many patients misdiagnosed for this same reason. This is exactly why we’re building InpharmD.
Isn't that what Up To Date is?
No shade, we love up to date; we just see ourselves as a complement.
How would you characterize the difference between InpharmD and UpToDate? Different features? Better execution?
This means they focus on the most common questions.
Public data/ our data shows point of care references like UTD can only answer ~1/2 of clinical questions.
We’re building our tool for the other 1/2.
Anyway, I'm not a doctor but one of those types that love looking into everything themselves. I'd be interested in having access to something like this for my personal research.
The reason every patient knows of the Physicians Desk Reference is because they believe their doc relies on the PDR.
In just 10 years I’ve seen the doctor go from being the top of the patient care hierarchy to the patient on top (not coincidentally DTC ads have blown up). We think there are a lot like you, and soon, we’ll make our tool publicly available.
If you want to test before that time- reach out! Ashish at InpharmD dot com
Agree with the first part enough that I wouldn't even want the near association :)
When Sherlock is older we may let him go by his middle name
A SAAS tool has a hard time with complex questions because healthcare data is messy and results will be imperfect.
Therefore we feel strongly about building a human - lead service thats optimized by software.
Every health system employs clinical pharmacists to manually do this today, so we use that as our stake in the ground, and so far we’re finding we’re way more efficient.
"I wasted $40k on a fantastic startup idea" https://news.ycombinator.com/item?id=25825917
I hope your project succeeds.
He went after the same problem, yes, but with a traditional SAAS approach.
We feel strongly that the only way to make this work is to have humans on our back end, thoughtfully automated with tech (vs the other way around).
Also, a newsfeed with common/ interesting questions has really helped since the experience gets better with more users.
When I was a medical student and resident on rounds, it was really stressful trying to do literature searches on pubmed and up-to-date (esp while taking care of active patient problems!) I often didn't have the time or bandwidth to comb through the papers/lists of care guidelines, and the abstracts often didn't give the clear story (or the relevant numbers!)
If I had had access to InpharmD, I could've 1) looked like a rockstar to my team, and 2) actually learned and understood the info I needed. :)
You can create a free trial account here and ask your question, all in 30 sec: https://www.inpharmd.com/provider_signup/new
Then if you like it, pester away!
1. What are your thoughts about partnering or selling to pharma / biotech / R&D orgs? Is there a potential value prop?
2. How does this compete with or complement existing clinical informatics and medical librarian capacity at academic medical centers? Or are they not the target market because they already pay salaries for humans to do these tasks? How do above entities relate to what the PharmDs do?
1. We tested this with Pfizer last year and found there was an opportunity to supplement existing med info teams that do the same thing.
But it’s tough to do two markets well at once, so we decided to focus on health systems for now.
We also find that in hospitals, everyone thinks they’re asking b unique questions, but they aren’t. We can be much cheaper vs their pharmacists and still make money. Pharma companies already have standard responses so it’s a totally different value prop.
2. We don’t really compete with clinical folks at hospitals, most will readily off load this to us, so they can spend more time on patient care. There are some that like to own this, and I totally get why, but we eventually win them over. As for medical librarians, they’re great for article requests but for complex clinical questions we think a clinical pharmacist is the right type of researcher.
This is pretty neat, hope it works out!
During vaccines, how have your pharmacists managed to do both? I’m sure it’s a lot on them
Am I answering what you asked ? Sorry if not!