Hacker News Reader

Comment by idiot900 | Hacker News Reader

tashfeens, I'm always skeptical about healthcare startups. A few questions: How accurate are your predictions, and how do you assess accuracy? (Do you have any prospective studies?) How is this better than the nanny reminders in current EMRs such as Epic? How does your platform deal with missing data when a chunk of a patient's visits are not transmitted to your platform?

Does the software parse free text, depend on someone taking the time to curate a list of "problems" or "diagnoses", or just look at lab values? What is the explanation given by the software when it makes a recommendation, other than deference to a statistical model?

Why would I, a physician, want to spend time or money on this rather than depend on my own sensibilities and training? (I am not going to watch the demo video on your website because I will not agree to an NDA.)

View on HN

I'm not tashfeens, but given that you have a CS degree and an MD, I'd be curious as to what makes you skeptical about healthcare startups. Or, conversely, what should a healthcare startup do to earn the respect of you and your colleagues?

Overall, make a case to truly be worth the time and effort.

Anything that second guesses my clinical judgment, and/or requires extra paperwork to do so, had better be able to show its justification in an intelligent manner. There is a lot involved in evaluating a patient that is not encoded in machine-readable form, so some percentage of things software reports is going to be wrong or pointless. (Trivial example: The K is 3.4. Is the patient hypokalemic?) The time spent ignoring nuisance notices may render the whole thing useless, even if there is useful information somewhere.

Have practicing physicians intimately involved in the UI design process, so the resulting product is usable to physicians and actually saves time. Billing software in disguise such as Epic has a terrible user experience and could be so much better.

There's more, but I have to head to work...

Thank you! That's helpful. If you'd ever like to talk further feel free to email me (email on my profile).

I can probably echo some similar thoughts that we heard when we talked with providers. I'm not tashfeens nor do I work for his company, but I run Varsa Health, a data analytics company for behavioral health providers. We "earned" the respect of clinicians to a certain degree by spending a lot of time (months) learning about their workflow before we started even writing code. My co-founder used to work on similar work in a research setting, so user experience was particularly important to us. I noticed you were working on Healthcare.gov - that's a massive undertaking and kudos to you for having the civic duty to help clean up the mess that it was. Would love to chat further about some of our lessons learned and share our experiences.

https://varsahealth.com/

This is great to hear. We're in a similar position, working on a startup for mental health providers. We spent a solid three months focused on user experience above all else and involving clinicians, case managers, and other end-users in the design process. It's a shame that more startups don't seek this from the get-go, especially those wanting to seek some kind of clinical validity.

The fact that it's not Epic is a huge plus in my book.

Eef: I've seen the horror that is Epic's 1980's terminal SQL nightmare backend. Lipstick on a dead, rotting windows-only pig running on citrix.

Fun fact: MUMPS predates C as a programming language. Sad that a company like that continues to do as well as it does, but then again it sells to admins, not doctors.

Reply on news.ycombinator.com