Module for COVID-19 #679
Comments
|
We don't think it would be very difficult, but there would be some limitations. This idea was also mentioned here: https://chat.fhir.org/#narrow/stream/179160-social/topic/Pandemic.20Hackathon My colleague @dehall had these thoughts:
|
|
That sounds promising. I'm hearing that the number of people infected doubles about every week. Could you use that information to determine what the static % chance of condition onset would be for a series of weeks rather than letting it grow continuously? |
|
Yes, you could. You can use Dates in the conditional logic if you want specific weeks. Number of infected doubles every 2 to 4 days depending on mitigation strategies. See https://medium.com/@tomaspueyo/coronavirus-act-today-or-people-will-die-f4d3d9cd99ca |
|
LOINC Codes - FHIR Observations SNOMED Codes - FHIR Conditions ICD-10 Codes |
|
A super minimal flow to get us started... Pre-Hospitalization Flow
Hospitalization Workflow
Characteristics From the Coronavirus Disease 2019 (COVID-19) Outbreak in China |
|
Let's be careful about not letting the R0 and Dates functionality block the rest of the flow. If we start with a flat percentage of population who are infected and run the population retroactively, people can seed the flow with whichever severity of a pandemic outbreak they chose to model. There's both outpatient condition modeling and inpatient hospital flow that we want to model and generate records for; not just modeling pandemic spread. |
|
Working in progress, if anyone wants to take it from here... TODO: The "Administer COVID-19 Test" diagnostic report needs work. I don't know what the values are supposed to look like. The inpatient admission needs a lot of work, and probably a loop in there... right now that is being abstracted away by the 1 - 21 day "Stay" delay state. |
|
Are you planning to simulate inpatient patient trajectories? Are the available aggregates good enough to emulate? |
|
I certainly think there would be benefit in doing so. Synthea can be used to generate both outpatient and inpatient populations, so I think there is benefit in both. The only limiting factor is the size of the workflow and understanding the model. Which is why COPD and Bronchitis are broken out into separate workflows, for example. Having worked in an OR and being familiar with scrub protocols and being connected with Agiliti (formerly Universal Hospital Services, formerly ABC Oxygen Tent Rental Company), I'm a little worried that I may get pulled into reserve operations in an ICU or ward. So I've been brushing up on ventilator mechanics: Merk Manual - Overview of Mechanical Ventilation And while we don't need to model every step of the process, it might behoove us to include respiratory rate, arterial oxygen saturation, , and PaCO2, as conditions for being put onto a ventilator.
Similarly, I would tentatively recommend adding |
|
For fever response, we want to use acetaminophen (RxNorm 161). WHO Now Officially Recommends to Avoid Taking Ibuprofen For COVID-19 Symptoms |
|
The most distinctive comorbidities of 32 non-survivors from a group of 52 intensive care unit patients with novel coronavirus disease 2019 (COVID-19) in the study by Xiaobo Yang and colleagues1 were cerebrovascular diseases (22%) and diabetes (22%). Another study2 included 1099 patients with confirmed COVID-19, of whom 173 had severe disease with comorbidities of hypertension (23·7%), diabetes mellitus (16·2%), coronary heart diseases (5·8%), and cerebrovascular disease (2·3%). In a third study,3 of 140 patients who were admitted to hospital with COVID-19, 30% had hypertension and 12% had diabetes." Are patients with hypertension and diabetes mellitus at increased risk for COVID-19 infection? And we want to add the following ICD10 codes somehow.
|
|
People with blood type A may be more vulnerable to coronavirus SNOMED-CT - Blood group A (finding) 882-1 - ABO and Rh group [Type] in Blood |
|
I'll take a stab at updating the module today. One reminder, Synthea does not support ICD10 codes... we use SNOMED, LOINC, and RxNorm. |
|
No ICD10???? How can that be? That's the oldest value set of them all! 400+ years old! In all seriousness, consider this whole thread a feature request for ICD10 support. It's an essential coding system... much more so than DSM-V, HCPC, MESH, etc. But we can do plenty with the SNOMED and LOINC codes in the meantime. And thank you for the help! |
|
ICD9 is public domain, so we could use that no trouble (with slight code change). We would have to look into ICD10 licensing from the WHO. |
|
Aaaaah. I see. They don't want people developing products with ICD10 without licensing it. As an interoperability product, can we create ICD9 and ICD10 containers as placeholders? Deliver the pipes, but not the content? I'm going to try to track somebody down from WHO that can speak to this. I have a pretty good rolodex, and am going to see what I can do. |
|
Yeah, COVID19 response would certainly qualify as non-commercial research license. I suppose the problem is SyntheticHealth also offers for-profit support of Synthea? Hmmm. http://apps.who.int/classifications/apps/icd/ClassificationDownloadNR/license.htm |
|
We don't offer for-profit support. Anyway, let's take the ICD10 question out of band, or another issue, so this thread can stay focused on COVID19. |
|
Right. I'll log a separate issue for ICD10 next week sometime. In the meantime, I'm wrapping up a deliverable, and am hoping to run the generator and load results into a HAPI server this weekend. |
|
Updated module. Needs testing and review. Lots (read: all) of observations and measurements are missing. But there are diagnosis, procedures, and medications. |
|
Another update. Still needs more work. |
|
I created a branch where I'll keep posting updates. It still needs a lot of work. Missing labs and observations, missing pathways, and probabilities that are just flat out wrong. |
|
So I'm relocated back to the family farm, and have my prior obligations completed and published. I'm focusing on COVID19 the rest of the week. Was just able to compile the COVID19 module, and so far so good. Getting a |
|
Huh. I've only been trying FHIR R4 and CSV, so I haven't seen that yet. |
|
I'll be checking R4 tomorrow. Just reporting things as I find them. :) I just managed to upload a generated dataset into HAPI, and am confirming the following queries. Looking great so far! # DSTU2 - Search for all testing encounters "Encounter for symptom (procedure)" ???
http://localhost:3100/baseDstu2/Encounter?type=185345009
# DSTU2 - Search for all conditions with "Cough (finding)"
http://localhost:3100/baseDstu2/Condition?code=49727002
# DSTU2 - Search for all conditions with "Dyspnea (finding)"
http://localhost:3100/baseDstu2/Condition?code=267036007
# DSTU2 - Search for all conditions with "Fever (finding)"
http://localhost:3100/baseDstu2/Condition?code=386661006
# DSTU2 - Search for all conditions with "Suspected COVID-19"
http://localhost:3100/baseDstu2/Condition?code=840544004
# DSTU2 - Search for all conditions with "Suspected COVID-19" in the month of March 2020
http://localhost:3100/baseDstu2/Condition?code=840544004&onset=ge2020-03-01&onset=le2020-03-31
# DSTU2 - Search for all conditions with "COVID-19"
http://localhost:3100/baseDstu2/Condition?code=840539006
# DSTU2 - Search for all conditions with "Pneumonia"
http://localhost:3100/baseDstu2/Condition?code=233604007
# DSTU2 - Search for medication admininstrations of "Hydroxychloroquine Sulfate 200 MG Oral Tablet"
http://localhost:3100/baseDstu2/MedicationAdministration?code=979092
# DSTU2 - Search for all procedures of type "Oxygen administration by mask (procedure)"
http://localhost:3100/baseDstu2/Procedure?code=371908008
#==================================================
# R4 - Search for all testing encounters "Encounter for symptom (procedure)" ???
http://localhost:3100/baseR4/Encounter?type=185345009
# R4 - Search for all conditions with "Cough (finding)"
http://localhost:3100/baseR4/Condition?code=49727002
# R4 - Search for all conditions with "Dyspnea (finding)"
http://localhost:3100/baseR4/Condition?code=267036007
# R4 - Search for all conditions with "Fever (finding)"
http://localhost:3100/baseR4/Condition?code=386661006
# R4 - Search for all conditions with "Suspected COVID-19"
http://localhost:3100/baseR4/Condition?code=840544004
# R4 - Search for all conditions with "Suspected COVID-19" in the month of March 2020
http://localhost:3100/baseR4/Condition?code=840544004&onset-date=ge2020-03-01&onset-date=le2020-03-31
# R4 - Search for all conditions with "COVID-19"
http://localhost:3100/baseR4/Condition?code=840539006
# R4 - Search for all conditions with "Pneumonia"
http://localhost:3100/baseR4/Condition?code=233604007
# R4 - Search for medication admininstrations of "Hydroxychloroquine Sulfate 200 MG Oral Tablet"
http://localhost:3100/baseR4/MedicationAdministration?code=979092
# R4 - Search for all procedures of type "Oxygen administration by mask (procedure)"
http://localhost:3100/baseR4/Procedure?code=371908008Update: Wasn't able to find any patients with "20 ML tocilizumab 20 MG/ML Injection". Looking through the raw files that were generated, it doesn't look like any MedicationAdministrations of tocilizumab were generated. |
|
Different countries/states are testing different percentages of the population. Finland is only doing minimal testing for the most vulnerable. In the situation where a country or state does not do much testing, we see many J06.9 Acute upper respiratory infection, unspecified. There are 1056 suspected/positive test covid cases U07.1/U07.2 and 47,258 J06.9 (Snomed 35207929) in the last 60 days. I would think the model could have some multiple ratio of U07.1 to J06.9 (Snomed 35207929) depending on percentage of population tested if that is possible. Weekends have lower diagnosis rates because of the slow onset of the disease. Here is something to show what I mean: https://share.geckoboard.com/dashboards/GFO7AU47LS4O4CJM/ |
|
@awatson1978 Is there some property setting to have the epidemic start in a particular month? |
|
@vpl-profess said:
The boxes are connected via the table based transition... there is just a bug in the module builder that does not connect the states. Sorry for the confusion. See the |
|
@Tracy-Mc It is a work in progress. I'll raise this question with some clinicians. |
|
@jawalonoski Great thankyou very much |
|
@jawalonoski just to add in the 'Infection pathway'. Compassionate Care does in fact have an option via 'recovery' to link back to 'take normal temperature'. It seems this is just not followed through in the admission pathway, as the only link from 'compassionate care' is 'death' |
|
@jawalonoski (same teams as Tracy-Mc !) Looking at the new infection.json file today it seems that things have disapeared. No blood test result and medication at all. Looking at github this file did not change for 7 days and we founded this blood test result yesterday ... totally lost. .. |
|
Some of the lab codes changed. I switched from a basic metabolic panel to a comprehensive metabolic panel, and added a few more things. The values should be in there, with different codes. Medications shouldn't have changed. If you have specific examples that might help debug the situation. |
|
@jawalonoski In the previous 'infection' pathway the compassionate care medication had a percentile lead to death and another lead to recovery. In the current 'admission' pathway, the tocilizumab medication order is in a similar place as expected however the only pathway from it leads only to death. The other link to recovery has been removed and linked instead to 'compassionate care check' above the medication order |
|
Ah, I was told (possibly erroneously, or I totally misunderstood) that if you are put on tocilizumab at that stage of covid19 you are basically going to die (hence "compassionate care"), so I removed the recovery pathway. I will get more clarification on that. |
|
I'm working with some clinicians who are going to take a deeper dive into the medications we're using and expand/alter them in different scenarios. I did get some more feedback on tocilizumab though, so I'm going to adjust that shortly. |
|
I'm working on Synthea again for the next week or two. Both on the Covid19 module and a proprietary pipeline related to Congestive Heart Failure. Turns out there is some overlap, so I'm starting there. Am going to propose some updates to the overall pipeline this weekend. Covid 19 - Congenital Heart Disease - Risk Factors
Other Risk Factors and Conditions
Other Risk Factors
Other Comorbidities (Stroke)
References
|
|
Proning & Positioning
Oxygen Therapy
UChicago Oxygen Therapy
Ventilator, Endoscopy, & Endotracheal Tubes
Ultraviolet Procedures
-UChicago Medicine doctors see 'truly remarkable' success using ventilator alternatives |
-UChicago Medicine begins clinical trial to explore blood plasma transfusions for COVID-19 patients |
|
|
Hey, |
|
Hi, I went and checked the Nasal discharge (64531003); I'm currently working on a pull request with it and a few other additions. |
|
Also, Dexamethasone is all in the British news this past week. |
|
And some articles on immunology:
LOINC Codes
|
|
We're already entering into some phase three vaccination trials, but they don't have LOINC or SNOMED codes assigned. However, we do have codes for MMR and Influenza. MMR Vaccine Appears to Confer Strong Protection from COVID-19 SNOMED CODES |
|
From Isreal, some research on blood clotting related to neutrophil counts, and the use of colchicine.
LOINC CODES |
|
In India, they seem to be getting a lot of mileage out of Favipiravir (Avigan, FabiFlu), but it's not listed in RxNorm. I did find a MESH entry for it though.
Can we code this up as a Substance? |
|
:sigh: This in from the disability community. Not altogether surprising, but sad nonetheless.
SNOMED CODES The Science Direct article has some statistical mortality rates, mostly based on age. |
|
Yes agree, dexamethasone should be included if not already. I have not looked for a while, however I thought it was already in the medication pathway, we are including in our COVID treatments
Tracy McClelland
Principal Medical Expert, UK&I
Healthcare & Life Sciences Industry
DXC Technology
M +44 7504685830
[cid:[email protected]]
Upcoming leave:
February 7th-11th
From: Abigail Watson <[email protected]>
Sent: 29 June 2020 03:27
To: synthetichealth/synthea <[email protected]>
Cc: McClelland, Tracy <[email protected]>; Mention <[email protected]>
Subject: Re: [synthetichealth/synthea] Module for COVID-19 (#679)
Also, Dexamethasone is all in the British news this past week.
* Coronavirus: Dexamethasone proves first life-saving drug<https://clicktime.symantec.com/3GMGHjsJBSGHmbsCTzPzvHJ7Vc?u=https%3A%2F%2Fwww.bbc.com%2Fnews%2Fhealth-53061281>
* Dexamethasone cuts covid-19 deaths<https://clicktime.symantec.com/32MZF1CnwFvmFBertjVbPyV7Vc?u=https%3A%2F%2Fwww.economist.com%2Fbritain%2F2020%2F06%2F20%2Fdexamethasone-cuts-covid-19-deaths>
* RECOVERY Trial<https://clicktime.symantec.com/3QnpUjbGKgUUSQsRjBT2mxd7Vc?u=https%3A%2F%2Fwww.recoverytrial.net%2F>
* Effect of Dexamethasone in Hospitalized Patients with COVID-19: Preliminary Report<https://clicktime.symantec.com/37ywQGQmuqFfVfcpHPQXyfa7Vc?u=https%3A%2F%2Fwww.medrxiv.org%2Fcontent%2F10.1101%2F2020.06.22.20137273v1>
* RxNorm 3264 - Dexamethasone<https://clicktime.symantec.com/33R2Mqd7sz8fRYtfFgnaR2J7Vc?u=http%3A%2F%2Fbioportal.bioontology.org%2Fontologies%2FRXNORM%3Fp%3Dclasses%26conceptid%3D3264>
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub<https://clicktime.symantec.com/3QMLJPoxpM8ekkXiu1e3sfs7Vc?u=https%3A%2F%2Fgithub.com%2Fsynthetichealth%2Fsynthea%2Fissues%2F679%23issuecomment-650867839>, or unsubscribe<https://clicktime.symantec.com/3FcyHDJ8v2HEn74YKs9ZaA97Vc?u=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAPDASE6GUCKTAYLBVJP5ZWDRY73Y3ANCNFSM4LGNAX7A>.
DXC Technology Company -- This message is transmitted to you by or on behalf of DXC Technology Company or one of its affiliates. It is intended exclusively for the addressee. The substance of this message, along with any attachments, may contain proprietary, confidential or privileged information or information that is otherwise legally exempt from disclosure. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient of this message, you are not authorized to read, print, retain, copy or disseminate any part of this message. If you have received this message in error, please destroy and delete all copies and notify the sender by return e-mail. Regardless of content, this e-mail shall not operate to bind DXC Technology Company or any of its affiliates to any order or other contract unless pursuant to explicit written agreement or government initiative expressly permitting the use of e-mail for such purpose.
|
The tables are located in |
|
Hi Jason, |
Immunization support -- not started. TBD. |
|
Since the COVID-19 module is now part of the |
|
Hello @jawalonoski @Tracy-Mc - has Dexamethasone been added to this module? I am running a ML model against Synthea patients who have been given a course of Dexamethasone. If not, is there a timeline for this to be added? |
No, it has not been added. No, there is no timeline. Feel free to add it yourself with the Module Builder https://synthetichealth.github.io/module-builder/#covid19/medications. We'll review a Pull Request if you or someone in the community submits one. |
We're wondering how hard it would be to generate some mock data for patients that may have COVID-19. As I understand it, this would require a new module to be created using the Module Builder tool. Our team works on the AI side though; we aren't medical professionals. Is there any interest from project contributors who do have the necessary background to take an initial stab at creating this module?
The text was updated successfully, but these errors were encountered: