Last revised: August 7, 2021
"An Experiment to Convert Structured Product Labels to Computable Prescribing Information"
Correspondence should be directed to: Dr. Allen Flynn at the University of Michigan (ajflynn@umich.edu).
The purpose of this work is to make FDA approved prescribing information not just available online but truly computer accessible.
Starting in 2005, the U.S. Food and Drug Administration (FDA) required pharmaceutical manufacturers to submit drug product prescribing information in XML format using a common XML document schema called the Structured Product Label (SPL). SPLs resulted in global online access to FDA-approved prescribing information via DailyMed from the U.S. National Library of Medicine (dailymed.nlm.nih.gov). People can read the content of SPLs using a computer browser but SPLs did not make that content computer accessible, let alone executable.
This project begins with an experiment to determine how the content in SPLs might be made computer accessible.
Here we provide initial experimental results from work to make the information in the SPL for remdesivir computer accessible.
The materials here are openly available for inspection, review, and comment. Send comments to ajflynn@umich.edu.
The first experiment for this project breaks down and then re-represents and reorganizes the FDA-approved prescribing information for Remdesivir. To do so, a question and answer pattern is used and the content generated is serialized as a JavaScript Object Notation (JSON) data object. Next, the JSON data object with its pairs of questions and corresponding answers (QA pairs) is then wrapped in a JavaScript function. This JavaScript function is then augmented with several other files to form a compound digital Knowledge Object (KO). Finally, when used with a tool called the KGrid Activator, the Remdesivir Label KO enables an application programming interface (API). Once the Remdesivir Label API is established then client applications running on computers can directly access the questions and answers that convey prescribing information for Remdesivir.
The files and work described in the paper are available for inspection and review here, including:
Link to File: PDF October 2020 Remdesivir
Link to File: XML October 2020 Remdesivir
An Excel file with Remdesivir Prescribing Information organized into 86 question-and-answer pairs (QA pairs)
Link to File: XL with 86 QA Pairs
Link to File: JSON with 86 QA Pairs
There are four files in the folder called "remdesivir-label-v1.0":
- an /src folder holding an index.js file (index.js)
- Deployment Description file (deployment.yaml)
- Service Description file (service.yaml)
- Knowledge Object metadata file (metadata.json)
Taken together, these four files and their file structure make a conformant Knowledge Object (KO). This Remdesivir Label KO meets the standards for structure and content of the open-source Knowledge Grid software being developed at the University of Michigan Medical School (www.kgrid.org).
KOs carry content that is computer accessible. In this case, that content is inside the index.js file. In addition, the Deployment Description carries machine-readable technical information about the runtime environment needed to run the index.js file, the Service Description carries a machine-readable description of the API, and the metadata file carries descriptive metadata about the Remdesivir Label KO.
This file can be downloaded at the link below and then imported into Postman. For more information about this, see section 4. INSTRUCTIONS FOR USING THE REMDESIVIR LABEL API.
https://github.com/kgrid-objects/fda-label-collection/releases/tag/Version1
We have deployed this API already. It is easy to try it! Just do this:
STEP 1. Copy the following four-line CURL command
curl -X 'POST' \
'https://activator.kgrid.org/js/fda_label_content/fda_label_content' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-d '0'
STEP 2. Using your browser, go to https://reqbin.com/curl
STEP 3. At REQBIN, paste the CURL command you copied above into the box on the left and click the 'Run' button
After clicking 'Run', the expected output when input is '0' includes ALL of the 86 QA pairs:
{ "result": { "QApair_1": { "QApair_label": "brand_name", "question_id": "b8408b213e84449dbb09fe9efc8fce44", "question": "What is its brand name?", "answer_id": "95dca543c2344635b15a90e3c2fac164", "answer": "VEKLURY", "answer_nature": "Unequivocal Answer" }, "QApair_2": { "QApair_label": "generic_name", "question_id": "ad47b354ee714cc8b7ad6ed73ca889c9", "question": "What is its generic name?", "answer_id": "0448074035454eacabeddaabb6777820", "answer": "remdesivir", "answer_nature": "Unequivocal Answer" }, ...STEP 4. Next, in REQBIN, change the '0' to a '1' in the last line of the curl command and run it again:
-d '1'
The expected ouput when input is '1' is a list of all question identifiers only:
"result": "[{\"question_id\":\"b8408b213e84449dbb09fe9efc8fce44\"},{\"question_id\":\"ad47b354ee714cc8b7ad6ed73ca889c9\"},{\"question_id\":\"b130fa66457849efb6272ec7deca93e1\"},{\"question_id\":\"7df59111730a49699f6864257c73738e\"},{\"question_id\":\"656d693bbe8f49b4b2f7f8527655636e\"},{\"question_id\":\"0109d944130344c2869ebc8e9329ac91\"}, ...STEP 5. Now, change the '1' to a '2' in the last line of the curl command and run it again on REQBIN:
-d '2'
This time, the expected ouput when input is '2' is a list of all question identifiers and their corresponding questions:
"result": "[{\"question_id\":\"b8408b213e84449dbb09fe9efc8fce44\",\"question\":\"What is its brand name?\"},{\"question_id\":\"ad47b354ee714cc8b7ad6ed73ca889c9\",\"question\":\"What is its generic name?\"},{\"question_id\":\"b130fa66457849efb6272ec7deca93e1\",\"question\":\"What drug products are included?\"},{\"question_id\":\"7df59111730a49699f6864257c73738e\",\"question\":\"What year was its initial U.S. approval?\"},{\"question_id\":\"656d693bbe8f49b4b2f7f8527655636e\",\"question\":\"What is its indication?\"},{\"question_id\":\"0109d944130344c2869ebc8e9329ac91\",\"question\":\"Where should it be administered?\"},{\"question_id\":\"a79effe7b17d4a778e13dc8858b6d225\",\"question\":\"What testing needs to be done before Initiating and during treatment?\"},...STEP 6. Next, change the '2' to a '3' in the last line of the curl command and run it once again on REQBIN:
-d '3'
The expected ouput when input is '3' is a list of all answer identifiers only:
"result": "[{\"answer_id\":\"95dca543c2344635b15a90e3c2fac164\"},{\"answer_id\":\"0448074035454eacabeddaabb6777820\"},{\"answer_id\":\"74d406dff7cd4774a538f1fe8bb90cdc\"},{\"answer_id\":\"bd10d1988c90413599e83481a46b1546\"},{\"answer_id\":\"7c5573a83ba54ac5b8b412988bd8f29c\"},{\"answer_id\":\"b2d9ea3ac32e4e2591342d7998a461d3\"},{\"answer_id\":\"3927118dce35411b90dda96143659f4f\"},{\"answer_id\":\"c4bff4466b1e480191d4738916c6d90c\"},...STEP 7. Now change the '3' to a '4' in the last line of the curl command and run it another time on REQBIN:
-d '4'
The expected ouput when input is '4' is a list of all answer identifiers and their corresponding answers:
"result": "[{\"answer_id\":\"95dca543c2344635b15a90e3c2fac164\",\"answer\":\"VEKLURY\"},{\"answer_id\":\"0448074035454eacabeddaabb6777820\",\"answer\":\"remdesivir\"},{\"answer_id\":\"74d406dff7cd4774a538f1fe8bb90cdc\",\"answer\":\"VEKLURY® (remdesivir) for injection, for intravenous use VEKLURY® (remdesivir) injection, for intravenous use\"},{\"answer_id\":\"bd10d1988c90413599e83481a46b1546\",\"answer\":\"2020\"},{\"answer_id\":\"7c5573a83ba54ac5b8b412988bd8f29c\",\"answer\":\"VEKLURY is indicated for adults and pediatric patients (12 years of age and older and weighing at least 40 kg) for the treatment of coronavirus disease 2019 (COVID-19) requiring hospitalization.\"},{\"answer_id\":\"b2d9ea3ac32e4e2591342d7998a461d3\",\"answer\":\"VEKLURY should only be administered in a hospital or in a healthcare setting capable of providing acute care comparable to inpatient hospital care.\"},{\"answer_id\":\"3927118dce35411b90dda96143659f4f\",\"answer\":\"Determine eGFR in all patients before starting VEKLURY and monitor while receiving VEKLURY as clinically appropriate. Perform hepatic laboratory testing in all patients before starting VEKLURY and while receiving VEKLURY as clinically appropriate.Determine prothrombin time in all patients before starting VEKLURY and monitor while receiving VEKLURY as clinically appropriate.\"},...STEP 8. Finally, using REQBIN, request a specific question and its answer by submitting a question identifier by using this input:
-d '"656d693bbe8f49b4b2f7f8527655636e"'
The output from this request gives the indication :
"result": "[{\"QApair_label\":\"indication\",\"question_id\":\"656d693bbe8f49b4b2f7f8527655636e\",\"question\":\"What is its indication?\",\"answer_id\":\"7c5573a83ba54ac5b8b412988bd8f29c\",\"answer\":\"VEKLURY is indicated for adults and pediatric patients (12 years of age and older and weighing at least 40 kg) for the treatment of coronavirus disease 2019 (COVID-19) requiring hospitalization.\",\"answer_nature\":\"Unequivocal Answer\"}]",Here is a link to a page where the API can be tried online in a different way using the Swagger Editor:
The link above should take you to a browser screen that looks like this:
To engage the API using the Swagger Editor, click on the green POST request button on the right to begin.
STEP 1. Download and install POSTMAN from https://www.postman.com/downloads/
STEP 2. Copy the following link:
https://raw.githubusercontent.com/kgrid-objects/fda-label-collection/main/fda-label-collection.postman_collection.json
This link points to the POSTMAN requests collection file that we created. Alternatively, you can download this file by right clicking HERE and select save link as; if you choose to download the file then stay on the
Filetab after you click the Import button (do not select theLinktab).
STEP 3. Run POSTMAN on your machine
STEP 4. Click on Import button as seen below
STEP 5. Select the Link tab, paste the copied link and click the Continue button
STEP 6. Expand the fda-label-collection and then click Example 1
STEP 7. Click on blue Send button which will initiate a POST request to https://activator-playground.herokuapp.com/js/fda_label_content/1.0/fda_label_content
By default, the body of the request should show “raw” JSON is selected and the answer id "ce519bb9d44f475bb4c3b0b8b5399fb6" (with quotes) on the first line
STEP 8. A successful response should return the “result” JSON object with information on the KO following it.
We included an
Example 2which after you click the blue Send button you should get back the following message:
{
"result": "\nInput error. Please try again.\n\t\t\n0 = send the whole remdesivir_label_object\t\t\n1 = send all of the question IDs as a list\t\t\n2 = send all of the question IDs and questions\t\t\n3 = send all of the answer IDs as a list\t\t\n4 = send all of the answer IDs and answers\t\t\nquestion/answer identifier = send the single object",
"info": ...
}
We have also provided an optional incomplete request called
Optional - Deploy the Remdesivir Label API on your own serverwhich is revelant to the section below. This optional API request is intended for folks who have implemented the Knowledge Grid Activator API Gateway Tool already.
Using Knowledge Grid technology - specifically the Knowledge Grid Activator API Gateway Tool - it is possible to download and deploy the Remdesivir Label to establish the Remdesivir Label API on any suitable server (including typical laptops).
To establish the API locally on a server, it will help if have prior experience with system administration and API deployment. We can only provide high-level instructions here.
The general steps to follow are:
- Download the KGrid Activator from its release site: https://github.com/kgrid/kgrid-activator/releases/tag/kgrid-activator-1.5.1
- Learn how to configure and use the KGrid Activator at: https://kgrid.org/kgrid-activator/#activator-quick-start
- Download the Remdesivir Label KO ZIP file release from this repo: https://github.com/kgrid-objects/fda-label-collection/releases
- First, unzip and then, once the KO is unzipped, activate the Remdesivir Label KO on your server using the KGrid Activator
- Engage the resulting Remdesivir Prescribing Information API wherever you have deployed it (see API use examples above)