Scary is the trend of open source maintainers sabotaging their own packages.
I work as a Developer Advocate in Mobile API Security and once you are working in a very sensitive space I would recommend you to pin all your dependencies in the mix.exs file to not let some sabotaged/compromised package get in without being noticed.
Does iTraX actually use Elixir (the language)? After following the links and looking around I haven’t seen any reference to Elixir (the language) being used at all. If so, it seems to be fairly well hidden.
It seem to me that the only similarity is the name of the company that developed the iTraX software – Elixir Software Ltd – having Elixir in it’s name.
Can anyone shed some light on this? Sorry, but I’m just a little skeptical here about the nature of this post.
A quick look around their website and I’m 99% sure it’s just a naming collision. Elixir Software would be a good company name for a business targeting pharma. This is from their hiring page:
We use a modern, open-source technology stack including Python, ReactJS, TypeScript, MySQL, graphQL, Neo4J. Everything we create is built for the cloud, using terraform & docker to orchestrate the 50+ AWS components. We efficiently deliver high quality software through agile development, collaborative working, regular code reviews and continuous integration.
I’m working in health care and I’m starting my Elixir journey. It’s quite a steep learning curve when you start from zero and gradually discover the whole of Elixir and OTP. I work on real-time data pipelines with high frequency medical records and it seems to me that Elixir is perfect for the job. I used the mllp and hl7 library by HCA to parse some HL7 streams and it was a delight compared to Mirth Connect for example. Right now I’m stuck because I want to stream the HL7 data to a Phoenix app but the dependencies on the telemetry package are incompatible between mllp and phoenix.
I feel like in health care we spend a lot of money for bad software and Elixir could improve a lot in this space.
Heya! Sorry about that! We neglected to update our dependencies there but will push up changes shortly.
Btw, you can override the dependencies in the mix files to force certain versions (and it should be fine since it looks like there’s no semantic versioning issue, they just moved from 0.x to 1.x to note a now stable API).
I’d love to hear more about what you’re doing or about any other issues/needs you’ve encountered with our libraries. Happy that you’ve found Elixir to be helpful!
Thank you very much for your message. As a workaround for the version issue, I cloned elixir-mllp as an app into my umbrella project and adjusted the version requirements and it worked fine.
I’m a medical doctor in anesthesia and intensive care and I build large databases of high frequency biometric data to train machine learning models and algorithms to improve patient care as well as real time pipelines to bring the algorithms to the clinician and apply them to real time data so they can help in decision making. Right now we are using mostly Kafka for the real time part but I feel Elixir could really shine.
I’m just exploring Elixir at the moment. Today I built a primitive phoenix dashboard, showing data received via HL7 ORU messages in real time using liveview. I will see how far I can push it. I’m using your HL7 package and phoenix-pubsub.
The learning curve is quite steep. I’m doing mostly Python otherwise. I’m starting to have deeply nested datastructures to feed the liveview and I’m struggling to see how I can manage them, like update with incoming messages and remove old data deep in the structure. I’m writing some convoluted recursive code that will be hard to maintain. I feel there must be better ways to go about this.
Thank you very much for your mllp and HL7 libraries. These are quite rare in open source since it’s a bit of nice and I haven’t seen any so well done as yours. Looking forward to seeing the FHIR library
I’ve worked in healthcare for a while and have one project I keep on meaning to build–based on my prior work at a company years ago.
I built a background job scraper with a RESTful API interface in rails, which was far heavier than I’dve wanted.
The API aggregated and correlated the FDC’s NDC, rxcui, and other datasets around the US’s prescription drug databases. It’s a little bit of a chore to cross-reference them, I feel like it’d be valuable to a range of med tech orgs.
The stack would be Phoenix for client docs, Oban for the big processing, and Absinthe for graphql. Maybe REST too.
Deployed onto RapidAPI for quick licensing with free limited access for evaluation. Open sourcing id parsing libraries and NDC API clients along the way.
Obviously I have a pretty clear vision for this, but of course time is the issue. Open to work with others and divide any revenue weighted by contributions! Just DM me!