Potential big data analytics for OpenMRS

Hi everyone,

I wanted to write to everyone about my job change being the community elected board member and also because I believe this could be a great benefit to our community.

I’ve taken a job with a data analytics startup in San Francisco called Lumiata. They are one of two companies used by Google for their health data analytics, so far they’ve created a differential diagnosis API where the EMR passes in de-identified patient data and the API outputs potential diagnoses.

I’m going to be Director of International Projects because they want to provide their tools in resource poor settings. If someone things their project could these tools or has ideas of other potential uses, for example for training purposes, data quality or other purposes, please get in touch because I’d love to talk.

5 Likes

Hi @jblaya, congratulations, and may I say that it’s great to have you living closer to us? it’ll greatly improve our chances of meeting in person :smile:

FYI, Just wanted to say that we (@maurya and I… well, mostly @maurya) have just begun work on supporting the OHDSI initiative for OpenMRS. To do this, we’re taking the OpenMRS data model, and converting it into OHDSI compliant format so that OHDSI apps can leverage OpenMRS data. It might be interesting to see how we can do the conversion from your own API’s to the OpenMRS API… and of course that leads to the ‘why do we keep having to translate from OpenMRS to other API systems, isn’t there a single method to do this?’ question. Not sure if this helps, but wanted to throw it out there…

Congrats on your new endeavor

I am starting to integrate openmrs with radiology stuff , and most of these reports are very wordy which makes analytics difficult .

We are still at the startup section for NLP , but there is def lots of opportunities for dealing with data and population health

Thanks Suranga. I looked at OHDSI and it looks really interesting. Would love to talk to either you or @maurya about the initiative. I’ve signed up for the conference in DC in Oct.

As far as the API, the first thing to be thankful for is a documented API which a lot of systems here in the US do not have. As far as translating it, I think you’re right about having to connect APIs and it not being a standard way, I have a feeling that we’re a long way away from standards for APIs. I know that Lumiata has to do data connections for every client individually.

@judy, what are the objectives of what you’re trying to learn from this? because we might be able to help.

Congrats @jblaya !! What do you envisage are the specifics for limited resource settings? I feel there are limited resources - human resources, tools - for analytics in the US or other high-income places

Good questions @sunbiz I’m in the process of what projects need and would love to hear from more projects about what they need/would like.

Currently what I’ve heard from the projects I’ve spoken with are the following ideas:

  1. Disease burden. Use EMR data to be able to calculate the true disease burden of a country or organization to then be able to deploy resources accordingly
  2. Resource optimization. Be able to deploy people and resources in a better way having analyzed needs.
  3. Medical training. Have clinical providers be trained within the EMR by having the system show them the probable diagnoses given the patient history and exam.
  4. Improve data quality. Show where data is missing or not being entered correctly.

Other ideas but which I haven’t found practical needs yet:

  1. Biosurveillance. Have the EMR or other system alert when there is a potential outbreak.
  2. Automated Triage. Have patients enter their information digitaly and have an automated triage system.

I think we have to consider how we are going to subsume similar concepts and take advantage of groupers to help with analytics. Jonathan Payne and I have been looking into this. Keep in the loop with any terminology-based discussions about this :smile:

Definitely Andy, the idea is to make something that is reusable by any OpenMRS implementation and part of that will be levering the dictionary and standards.

Joaquín

I just wanted to chime in and say that this topic really interests me. I’m sorry I can’t be more helpful at this point, but if you’re ever looking for some developer input, please include me :smile:.