The COVID-19 CDSS solution – technical architecture overview
Our COVID-19 CDSS is designed as a scalable cloud hosted service that can be accessed and used in a variety of flexible ways. It is designed to work with standards compliant clinical data that are coded using SNOMED CT. The clinical data sets are custom designed using archetypes from the openEHR COVID-19 project and available freely. This ensures that the clinical data gathered remains semantically interoperable always. This has been done with the aim of making the data available easily for any research that follows proper ethical and privacy guidelines.
Flexible ways to access the use the CDSS service
At a primary level the service can be accessed and used through the custom mobile & web applications that are created as part of the project. These applications are being developed and hosted by the EHRC@IIITB team. We plan to have applications available for both Android and iPhone platforms, apart from the cross platform web application. To guarantee privacy, we do not collect any Personally Identifiable Information(PII) through these applications. They are designed as a supporting tool for doctors and other healthcare workers and not as a full fledged EHR solution. You can use them in tandem with any other digital health solution that you may be using.
At a secondary level, the solution provides REST APIs to enable any other applications to easily integrate the service into their offerings. By using the basic set of APIs, you can make this service available to all your users from within your application. Again the CDSS service does not require any PII to provide the service. However, to enable continuing monitoring, treatment and research, deeper integration is also possible with the use of de-identified personal data that includes patient ID from the originating application. This can be used to connect data from both systems, based on appropriate consent from the person, should such a requirement arise in the future.
Main components that make up the COVID-19 CDSS service
The common public REST APIs of the service, that are used by own and third party applications are exposed by the CDSS Application Server. A developer friendly Openapi standards based documentation of the CDSS Application Server REST APIs is available for your reference. This modular app server is built using Spring Boot and can be extended to include more functionality if required. It implements the application specific business logic and exposes simplified APIs for easy integration. The server abstracts the complex resources of the back end into simple key-value JSON for the upstream applications. It also structures and codes the incoming data before committing to the EHR repository at the back end.
The CDSS server has been implemented using Spring boot framework as a standalone service that can be deployed flexibly. This service uses Drools engine for implementing the flexible rules required to build a knowledge based CDSS. These rules can be evolved and managed dynamically using DRL syntax. We have designed this service to be flexible and capable of hosting multiple protocols at any time. Applications can choose which protocol to use while requesting decision support from the service. Currently the protocols and associated rules are being designed by a panel of doctors from the DHIndia association. A parallel research project to create a Bayesian Inference Network is being pursued by a team of volunteers. If and when field data becomes available, we plan to enhance the solution using ML/AI tools.
EHR.Network, our EHR as a Platform provides the core services that hold the entire solution together. This CDSS solution uses the Demographics, EHR and Auth services from EHR.Network. The entire solution uses the EHR platform for all of its data storage and security requirements as the remaining components have been designed without any local data repositories. The platform itself is compliant to standards and follows the Privacy by Design Principle. It has also been audited independently to conform to the OWASP top10 security guidelines.
This solution has been architected to be extensible into actual patient treatment flows, should the need arise at any point of time. Given the EHR.Network at its core, we can quickly ramp up this solution to handle clinical other data sets that may be required by such solutions.
- Project page – https://dhindia.org/2020/05/26/clinical-decision-support-system-for-covid-19-screening/
- API documentation for developers – https://cdss.ehr.network/covid/swagger-ui.html
- Project related post – https://ehr.network/clinical-decision-support-solution-for-covid-19-on-ehr-network/
- Post on EHRC website – https://ehrc.iiitb.ac.in/project/clinical-decision-support-solution-cdss
- Post on DHIndia website – https://dhindia.org/2020/05/14/dhindia-initiative-to-create-a-clinical-decision-support-solution-for-covid-19-screening