We have 95 responses (1 might be a student). If everyone is in agreement I can update and smoothen the abstract and one page report. This was done largely because it didn’t seem like we would have enough participants to create a panel. Input from others please!
1 Like
I was able to delete the student response so there are now 94 responses. I plan to send out the results to the folks who took the survey. The Powerpoint did include the EHR vendor used. Here is the link to the PowerPoint slides. When I get a chance I’ll do a few cross tabulations. Please give me your feedback
To my IT colleagues Tony and Art. Have you noticed the sudden interest in blockchain technologies in healthcare by ONC and others? Any idea how we might leverage this in a grant? We could tie together a chain of LibreHealth EHRs in a region, I suppose. Please educate me 
Wow. I missed the Forum entry already on blockchain technology
Yes, for sure and I have a colleague that just got a grant for that kind of work that is interested in joining LibreHealthIO, it so happens.
And there’s this: http://www.emrandhipaa.com/katherine/2017/01/16/ibm-watson-partners-with-fda-on-blockchain-driven-health-sharing/
Sunbiz pointed out the 15 ONC white papers that were published about blockchain technology and healthcare. One of the articles dealing with EHRs I have just printed out and need to review. Seems like this would be a good avenue for funding as it appears high on the ONC radar screen
I had a very productive meeting with our research team today. There are several pots of money we could use for development of an educational prototype using NHANES data. In case I didn’t post it already, here is a link to the 60 data tables that would need to be present in LibreHealth EHR, in order to receive the data from NHANES. To Tony and Art: now many man hours would it take to make this happen and what is a ball park $ estimate? Can this be achieved by this summer?
What I am thinking, but I need input from others is that in phase I we would build an educational prototype and web host it. Perhaps we can get 5 universities to use it and evaluate it. Clearly, I can use it in the 3 courses I teach. BTW, I have private server space I have paid for if needed.
We would at the same time, seek funding to do more than just education with LibreHealth EHR. We could use it as a research platform, database management platform, blockchain prototype, integrate it with mobile platform for nurses in the field, embed data analytics and so forth. Our research department is looking for NSF and NIH grants but also wonder about a SBIR grant. I made it clear our partner would be IUPUI, so I want to be sure we get input from them. Very exciting!!
1 Like
So … very quick look - their definition for “table” is clearly not the database definition. For example: BPXSY2 blood pressure systolic BPXDI2 blood pressure diastolic BPXCHR heart rate BMXWT weight (kg) BMXHT height (cm) BMXBMI BMI BMXWAIST waist circumference (cm) BMXTRI triceps skinfold (mm)
These 8 “Tables” are in the EMR as one table called Vitals. Do they really mean that they have a spreadsheet containing some number of systolic blood pressures for some number of patients and a separate spreadsheet for diastolic for the same list of patients?
In my view this is 5 “database tables” and the “Questions” one is new, everything else already exists in the EHR as a structure.
@tony Most of the questions are history questions and are in the EHR as the family history tab. My 2 cents for what it is worth, 85% of the fields are there , The data will just have to be massaged into the correct spot. Summer deadline doable IMHO.
There are a few more issues to address:
- Art seemed to have the ability to generate fake names and addresses. That would be great to match up with the existing demographics to make patients seem more realistic
- We need to figure out a way to import the vital signs into the single table “Vitals”
- In the section labelled “Questionnaire” they simply ask the patient if they have ever been told they have hypertension, diabetes, etc. There is no ICD coding in NHANES so what I thought we would do (which is only semi-kosher) is the take those who answer yes to e.g. hypertension and label them with the ICD-10 code for hypertension. We might list 3 pertinent codes for each disease entity to give it some variety. Does this make sense?
- We need to discuss how we might host 5 instances of LibreHealth EHR and have 5 schools test it.
- We also need to discuss in the future how to install the EHR and patients automatically for those in the future who use the components on GitHub
The spreadsheet now has ICD-10 codes for the common diseases mentioned. The first letter is frequently a capital I and not a 1 (looks similar)
Getting the data will not be any harder than any data conversion we’ve done in the past. Those typically cost a customer between $5-$10K depending on the complexity.
Using Arts tool to generate Patient records is good, just have to figure out a good random ID connector tool to the NHANES data set.
@r0bby can help us come up with a good, repeatable and reset-able deployment model.
Just let me know when you need me and I’ll be happy to.
That’s the kind of information we need. If we can guestimate $10K for the conversion, how much more for auto-loading capability? I think we are in the ballpark where we can use immediately available funds to get the prototype up and running. We will probably need a contract with statement of work, deliverables, timeline and so forth. I don’t know if Sunbiz would like to add a grad student to the mix. Not sure if UWF has any ready for a project.
By the way, the NHANES files are in a SAS transport format. Here is the link to the SAS viewer that will allow you to convert to CSV
1 Like
Talked to Michael @downey today and we both agree that this is totally doable through the organization, we have some formalities to setup, like a project lead and committee and some discussions with Software Freedom Conservancy on how to handle the funds correctly (both internally and with vendors/contractors). But we can do that quickly.
We (the project) can host the 5 sites as demonstration projects and then, when the universities want to bring it into their environments we can assist (as a community) or recommend a vendor from the community as needed.
So the hosting side of the house is purely a donation to the project as we have the resource already. perhaps Michael can take a stab at the value of that.
–tony
Sounds good. As soon as you work out details on your end let me know and I will see how the University of West Florida will initiate a contract. The 5 sites I pulled out of the air, by the way so could be 3 or whatever number
We have some resources from rackspace plus some space left on our proxmox infrastructure. Let me know in advance when you need by making a support ticket.
While it may seem a little early to talk about adding functionality to the existing LibreHealth EHR for educational purposes, the reality is that when we seek serious grant money we need to have a solid proposal based on need and feasibility. Therefore, it would be great if the development team would comment on the following
- As mentioned before we created an open source diabetes registry with components stored on GitHub. How might this be integrated with the EHR? This is way more than just creating a patient list of all of the type 2 diabetics in a practice. This would be a wonderful asset for the EHR for clinicians and academicians.
- Another very innovative idea would be to have the ability to run analytics on patients we add to the EHR in phase I. This would require grant money for programming, etc. The software we would use is the open source program known as WEKA that not only has an API, but in a standalone application you can link to a MySQL database. I have a vision of how this might work, but lack the technical know how to accomplish this. Is there anyone on the development team who would work with me on a “reality check” to see how we might accomplish this, so we could state this as an objective for a grant to enhance LibreHealth. No EHR comes with an analytics package!
I don’t know much about WEKA or JAVA myself but there are plenty of folks on this list that are Java experts.
The last data mining / ETL tool I used for OpenEMR was Pentaho and I have all the code and configurations for that as a base line on the oemr501c3 web site and we could fork that if it’s of interest: https://github.com/oemr501c3/openemr-insights
As to the registry, it’s also Java (a servlet) I’ve integrated those into the EHR before. Not sure how the EHR would feed or be fed from that system, I guess I’d need to see a demo or something. Since it’s a servlet it requires TomCat server to run alongside apache2. Not a big deal just added configuration burden. @r0bby might be able to containerize it to ease the pain
As I understand it Pentaho machine learning is really WEKA. WEKA is open source and has a large development community so we can get their support, if needed. They frequently state it can be integrated with other applications. This might end up an important selling point if we can offer a seamless way to run simple predictive analytics within LibreHealth EHR using WEKA. This might open up new grant possibilities. My Research Department is investigating research grants and hope IUPUI is doing the same so we can combine forces