Today, I am excited to announce the availability of the new Streams Healthcare Analytics Platform project on Github.
In the last few months, we have been busy talking to our business partners and customers about their requirements and how we can help them more easily build real-time healthcare analytics applications. We discussed creating a healthcare toolkit, where we will provide operators and functions to help build the application. We have also brainstormed on ideas of creating turnkey solutions for all healthcare application needs. Neither of the solutions seem to fit well with our healthcare customers.
With a toolkit, instead of focusing on developing useful healthcare analytics, customers are still distracted with the tasks of writing a Streams application using the toolkit’s operators and functions. Having a single turnkey solution does not seem to fit well either, as each of the customers can be trying to solve very different and unique problems. It’s hard to design a one-size-fits-all solution.
Streams Healthcare Analytics Platform
The goal of the Streams Healthcare Analytics Platform is to enable our users to build real-time healthcare monitoring and analytics applications with little or no coding at all. The platform will handle all common plumbing and infrastructure tasks related to a healthcare application, allowing our users to be more focused with developing life-saving healthcare analytics.
The platform is designed to employ the microservice architecture as proposed by Dan Debrunner in this post here. A service is a small Streams application created to solve some specific needs in a healthcare application. For example, a service can be provided to ingest data from a common bedside device integrator. An application is a composition of one or more services. Users can dynamically mix and match different services to create an application that suits their unique requirements. I will be providing more details to discuss microservice design in the healthcare platform in a future post.
Our Initial Contribution
This week, we have contributed an initial set of services on Github. In this contribution, we have provided a set of ingest services to help customers with the problem of ingesting patient’s vital and waveform data into a Streams application. We have also provided an ECG analytics service for detecting R-Peak in an ECG signal. To demonstrate that this architecture works in the healthcare setting, we have also contributed a real-time ECG monitoring application, using the Physionet Ingest Service, BioSPPy R-Peak Detect Algorithms and Python Jupyter Notebook.
You can try out this sample from here.
Design and Roadmap
If you are interested in the general design and roadmap of the Streams Healthcare Analytics Platform, check out this page.
We are looking for feedback and requirements for the platform. We also welcome contributions in this space. For any feedback, please submit an issue to the Github project here.