Skip to Main Content

AI based COVID testing platform

Fully AI based image recognition of variety of Rapid Antibody and Antigen and PCR tests from Roche, Abbott, SD Biosensor and SensingSelf which are are top providers of the testing solutions.

Also, integrated national pulse dashboard to evaluate, project and isolate high risk areas in terms of clustering and identifying pooling sites.

Platform is based on Mobile phone iOS and Android app.

edited on Nov 19, 2020 by keyur Patel
Public (8)
You will need to login to post a comment
Andrea Pitkus Nov 16, 2020

What's meant by pooling sites? Are you trying to identify where pooled COVID testing might be performed? If so, surveillance, diagnostic or screening pooled testing? How will the pooling aspect change when disease prevalence changes in a population?

keyur Patel Nov 17, 2020

Ye.s we are mapping pooled COVID testing cells and subcells for surveillance and predictive modeling using AI and machine learning. We have identified several future programs for disease management (eg. Common Influenza, Diabetes, Hypertension) and in those areas data can be collected/mapped and leveraged for future pulse of collective health.

Andrea Pitkus Nov 18, 2020

I'm not sure I'm following. What is meant by a pooled COVID testing cell? So you may know that surveillance testing is out of scope for the HHS/ELR reporting for COVID-19 per CMS. (Also why I was trying to clarify if you are handling data from pooled screening tests or pooled diagnostic tests as those would qualify for the reporting requirements on the homepage of the Design a thon). Can you provide further clarification?

keyur Patel Nov 24, 2020

Pooled cell is defined by zonal sites for testing and tracing. I understand it is out of scope for HHS/ELR. We have API which allows for integration with Pooled Screening and Pooled Diagnostics tests.

Andrea Pitkus Nov 16, 2020

How will pre-analytical, analytical and post-analytical laboratory testing aspects play into your approach, as well as CLIA/state laboratory regulations? see https://www.ecfr.gov/cgi-bin/text-idx?SID=124...e42.5.493_11291

keyur Patel Nov 17, 2020

We have compliance templates for data structures for eFCR related approaches and suggestions. Currently platform is open for modification for accommodation of regulatory nuances that may not be covered.

Andrea Pitkus Nov 16, 2020

Also interested in how HHS COVID reporting requirements would work with the model (AOEs, UDI info, etc.) ? Assuming LOINC, SCT and other maps will be used behind the scenes? Who's the audience for this platform (individual patients with the apps or performers of the testing with the kits in the first paragraph or downstream agencies with the dashboard?)

keyur Patel Nov 17, 2020

Correct. LOINC, SCT and other maps used behind the scenes. This platform is designed to have multiple views, Dashboard can be accessed, leveraged and used as National Dashboard, State Dashboard, Community Dashboard, Provider Platform, Clinical Platform etc App can also be used by patients in addition to the providers, health workers and downstream agencies.

Andrea Pitkus Nov 18, 2020

Given you indicate you support a variety of platforms and today the first patient performed home test has a EUA, is that supported? If so, how will patient AOE, UDI information be captured by your mobile platform, especially where the different IVD vendor kits may be utilized by the same or different patients given encoding and other details will vary by IVD vendor? In this use case, pooled methods would not be deployed.

If this were used with POCT with health professionals, curious how they would have the correct encoding/data required by ELR/HHS for COVID reporting too. Like the app link above to see the GUI/Design, but missing how results captured on a Smartphone are transmitted to public health (ELR/HHS).

In these cases, the dashboard would not be utilized in this point in the work/information flow, but may be something further downstream that is used to analyze/process these data aggregates.

keyur Patel Nov 19, 2020

We are today looking at Lucira Health platform for data capture. Our system relies on 'image processing" which cannot be used in Lucira's case, since it is electronic capture. But our API if integrated with their device can autodetect the results on flag and upload it to the third party systems from the cloud. In case of different IVD vendor kits, we use QR code and other relational image as well as manual detection / input process to differentiate the vendor codes. All data edit/and archives are rule based and POCT centers have web access to the data for sophisitication of capture/modify functions for ELR/HHS reporting. Dashboard use is intended for downstream use, which is correct assumption.

Andrea Pitkus Nov 23, 2020

There's also different combinations of the following:
1. Medical Laboratory Scientist performed in lab testing using traditional EHR/LIS systems for tracking, entry (manual/interface) and laboratory managed POCT
2. Non laboratory performed testing by non laboratory/other health professionals. Provider needs to order, but may be sole provider blanket order for university/national guard, pharmacy, etc setting. AOEs may be provided by patient in webportal and married with results and reported to public health by performer system.
3. Provider orders and sends test kit to patient. Patient performs testing and those results need to be scanned/detected/entered in a system along with their demographics, etc. With Lucira, results goto provider who then needs to report to public health via eCR (electronic case reporting) since lab is not involved (ELR doesn't appear to apply).
4. Patient/provider orders test kit which is sent to patient, patient collects specimen and sends to performing lab (i.e. Lab Corp, Quest, or CVS/Walgreens drive through) and lab performs testing, reports to public health, provider and patient.
5. Patient purchases test kit and performs test at home. Results may be read/captured by smartphone along with patient demographics. These need to be reported to physician and public health but unclear if ELR pathway or eCR pathway (for physician reporting which is different than lab reporting in terms of requirements/laws)

keyur Patel Nov 24, 2020

We do differentiate Test Types, we only need to do preselection of Type of Testing Service and Diagnostic method Template can help create single view for multiple venues of testing setups. 1 thru 5 are all modeled differently.

Andrea Pitkus Nov 16, 2020

Trying to understand if there are preconditions met by other software. (In general) 1. How to collect AOEs, from ordering provider/patient/specimen collector 2. Integrate into app/LIS or other information source for patient to be married to results of IVD test device/system 3. All transmitted to public health (ELR) 4. All transmitted to HHS (may be met by 3).

Will your approach support all 4 or only portions of above? Looks like your steps 0-9 mostly meet #1 above. #2 above appears to be mostly met with steps 13-14, but unclear how terminology maps are applied to results and result values? Unclear how #3 and #4 above are met in the app, but further downstream data appears in the analytics provided.

keyur Patel Nov 17, 2020

1. In terms of pre-conditions for third party software, we provide standard Open Data API's to either share or retrieve data based on credentials and compliance rules. Any ordering/patient/specimen collector data structure can be integrated in the template form or modified for organizational/provider usage. 2. Integrating app./LIS into information sources of IVD test system can be done at API level. We have structure to integrate Abbott or other third party api's. 3 and 4 / Transmission is rule based and security compliance based protocol to be defined by implementor.

Andrea Pitkus Nov 18, 2020

With your point 2, it's usually the opposite in traditional global laboratory workflows. 1. The laboratory determines which tests/kits are on their menu and builds orders and results in their LIS for them so the data can be captured, stored, transmitted. 2. These test catalog orders/CLIA specimen collection manual test info are shared with all ordering providers in compliance with CLIA regulations. 3. EHRs usually build (instead of consuming) these orders. Part of this build are the AOEs described in the HHS/ELR COVID-19 reporting specifications. 4. EHRs on the patient instance order present the AOEs to the provider/specimen collector such as drive collection site/patient if pre visit scheduling app/website to provide the responses required by regulations. 5. These AOEs, responses are sent with the order and specimen to the performing laboratory. 6. The laboratory has IVD test systems either interfaced to the LIS so results generated on them populate the LIS or with POCT results are manually entered into the LIS (or other app) 7. Results are married with AOEs, provider info, etc. and sent to public health ELR as required by law. 8 public health removes PHI and sends on to CDC completing HHS COVID reporting requirements.

Regarding your points in 3 & 4, do the specifications from the implementor include the HHS/ELR requirements or are they potentially different?

Your approach has potential, but since not much detail above, trying to understand if it also meets all the federal requirements or only a portion of them/portion of workflow/processes?

keyur Patel Nov 17, 2020

What is presented in the deck is sectional view / summary of the app. There are several other aspects not presented to relate to your #3 and #4 for collecting/sharing/tagging downstream data.

Andrea Pitkus Nov 18, 2020

Thank you. It looks like mostly the app collections patient info/AOEs, as well as provides a nice interface for using a smartphone to collect patient performed COVID testing with your IVD kits. The app provides the patient with a passport indicator within the app that can be shown for work/school needs. Hoping these patient data can be mapped, and also submitted to providers for clinical decision making and public health for contact tracing, etc and report similar to laboratories under CARES Act. It would allow greater deployment of these testing paradigms, while maintaining the same quality of data collection/requirements seen in healthcare settings.

This approach may be more important as hospitals/health settings fill beyond capacity, there are shortages of health professionals to care for these patients, perform testing, etc. Yet allow public health and government agencies to be better informed of case/public health needs, outbreaks/spreads in settings/populations to try to mitigate and respond more effectively. It may be needed as COVID care expands to non traditional settings too.

keyur Patel Nov 19, 2020

Absolutely right. Also we have more self testing platforms using saliva and urine and blood so we have built the app for future deployments to further capture downstream influences on containment of flu zones, diabetic communities etc. It is intended to create larger platform on and ongoing basis so pulse of communities can be taken and mapped to actual reality of disease management.

keyur Patel Nov 17, 2020

Also see our S4 Multi virus testing also supported on same platform for collection of disease detection data and results. We anticipate more testing platforms will be emerging in coming times to expand beyond COVID.

keyur Patel Nov 26, 2020

Dear Folks, Total Samples uploaded for Machine Learning crossed 45 million datasets today. Total predictive models created at 1.2 billion sets. On track to add another 20 million datasets by Dec.2020

sonemarshall91 Sep 28, 2022

Also, integrated national pulse dashboard to evaluate, project and isolate high risk areas in terms of clustering and identifying pooling sites.
mobile diesel mechanic

sonemarshall91 Sep 28, 2022

Also, integrated national pulse dashboard to evaluate, project and isolate high risk areas in terms of clustering and identifying pooling sites.
mobile diesel mechanic