Hello All,
I want to create a model where a user can easily find drugs associated with one or more medical condition(s). Can anyone help me with the databases available online?
Thanks in advance
1 Like
Link below is for a repository of queries against the vocabulary and CDM. Maybe this will help
OK. What about inclusion of off-label use?
Or do you need only official (per FDA (/EMA/others) label)?
Josh Denny published a knowledge base for that few years back.
Please refer to the QueryLibrary for the latest work in this area:
You should pull from Master for the latest, most stable code. Installation instructions and setup are available on the github page.
Hi,
I have been working on a project trying to build these exact links but ended up stalling because it was hard to automatically validate the results in a meaningful way. We used co-occurrence statistics to generate our links. Two resources I used for comparisons:
MEDI (the Josh Denny built resource mentioned) and the validation paper for AMIA: https://www.vumc.org/cpm/cpm-blog/medi-ensemble-medication-indication-resource-0
COHD: A database of concept co-occurrence statistics (for all concepts) built at Columbia http://cohd.io/
alt link is https://data.ohdsi.org/QueryLibrary/
D16 seems to assume you purchased First Databank.
The plan B in the query seems to be NDF-RT
Well, NDF-RT is sunsetted and replaced by MED RT (indirectly via RxNorm inclusion?)
Is MED RT now being core part of OMOP Vocab?
What about RxNorm extension drugs. (and after growth of drugs within SNOMED CT). I expect those are not covered by MED RT.
This subject keeps coming up, and folks are looking for Condition concepts that would be associated with Drug concept. Doesn’t really work that well.
What’s your use case, @VISHAKHA_TIWARI?
Hello @Christian_Reich
I’m trying to create a model which can help users to find the drugs for one or more medical conditions. for example, when a user search for MS the system should Mayzent drug as it is for the Multiple Sclerosis. for this I would need ICD-10 CM and NDC mapping, but I don’t know how to achieve this.
I understand. Except - why do you need that? You might want it:
- To support patient care, so the doc looking at a patient with MS will see Mayzent. DON’T! The OMOP CDM is not designed for patient care. We are studying populations in an observational setting (patients all in the past).
- To support cohort construction. This would be useful, but as I explained in the other post it doesn’t work, because the way patient are diagnosed is at a different level of granularity than treatment assignment is done. As in the example of the itching cream. The indication will be “itch”. That is not a diagnosis. There could be a gazillion reason why there is an itch.
Also, generally we are trying to create associations between drugs and diseases, i.e. it is the output of our work. You want that as the input.
We have created this map between AIs and Conditions. There will be a major update in early spring.
Our users use this information to create their digital health records. Yes, it is not a diagnosis but it is helpful for the user to keep a record of what medication was taken and the best reason they can input. All of us do this but never create a record. There are a lot more benefits to the patient with such mapped data.