Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Open-source patient model tops industry standard (umich.edu)
69 points by rbanffy on Feb 26, 2022 | hide | past | favorite | 15 comments


I would be surprised if new tech like this really sees meaningful adoption. This is the same industry that can't even get rid of MUMPS[1], the worst hybrid database/language atrocity ever.

[1] https://en.wikipedia.org/wiki/MUMPS


MUMPS is a really fascinating system to study. It has some good ideas (integrating data storage and the application implicitly with operating system management, sort of a protofilesystem) and some really bad ones (every datatype is basically just a string >_>). It's very clearly a first generation system but the fact that it has stuck around in its applications a bit longer than it should've is, at the very least, a point of interest.

FWIW, healthcare does integrate new tech relatively rapidly if it provides a sea-change in patient care and, primarily and somewhat less fortunately, profitability.


> FWIW, healthcare does integrate new tech relatively rapidly if it provides a sea-change in patient care and, primarily and somewhat less fortunately, profitability.

One of the issues when it comes to healthcare, unfortunately, is that inefficiency in technology is often a driver of profitability. Removing inefficiencies in certain sectors has a noticeable downwards effect on profits, but a less noticeable effect on improving patient outcomes. Overall care may be increased, but generally over much longer and harder to measure terms.


> some really bad ones (every datatype is basically just a string >_>).

Tcl says hello. :)

* https://wiki.tcl-lang.org/page/everything+is+a+string

* https://news.ycombinator.com/item?id=21377517


> every datatype is basically just a string

Wow! Like they invented nosql ahead of its time


There were a wide range of "NoSQL" databases in production use long before SQL (or even the relational model) was ever invented.


And many of those have data types.


This isn't true.

Models get replaced all the time - they usually only have a single integration point and little connection to the business of healthcare.

MUMPS will never be replaced because there's no incentive for the industry to move away from it.


>>MUMPS will never be replaced ...

Never is a very long time. Things change.


> Never is a very long time. Things change.

Yes that's fair.


The open source model works better than epic’s because it’s actually trying to solve a problem, rather than making a minimal feature for marketing purposes


Yep, it's funny how well things work when they are designed to solve a problem rather than to create profit for a company.


The nice thing is that models like this can now be plugged into Epic (and other EHRs) as SMART on FHIR apps.


I am in process of digitizing my paper HR – I am doing it as a webpage with semantic data attached as RDF tags. Do you think there would be a way to make it compatible with said apps. What does those apps provide please?


Here's the specification. A SMART on FHIR app can do pretty much anything like clinical decision support or prior authorization or drawing graphs as long as the hosting EHR supports it.

https://docs.smarthealthit.org/

In theory you could convert your paper health record into HL7 FHIR format and then post it on a server like HAPI so as to make it available to SMART on FHIR apps. But in practice due to security concerns and technical complexity it's going to be difficult to get your doctor's IT department to connect their EHR to some random patient's server. You're better off asking them first what data formats or transport protocols they support first.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: