By Jignesh Desai, Partner Solutions Architect at AWS
By Dr. Suman De, Head of Government Healthcare Analytics at Infosys Public Services
By Rajib Deb, Associate Vice President at Infosys
Infosys Public Services (IPS) and Amazon Web Services (AWS) have collaborated to develop Infosys Health Insights Platform, an advanced data science platform that helps government healthcare agencies turn multi-source, multi-format data into insights and actions.
Its goal is to deliver proactive and targeted interventions to the right people at the right time. This can improve care experience and outcomes across areas like population health management, epidemic management, behavioral health, and substance abuse.
Infosys is an AWS Partner Network (APN) Premier Consulting Partner and AWS Managed Service Provider. They have AWS Competencies in Data & Analytics, DevOps, Digital Customer Experience, and more.
In this post, we will walk you through the path Infosys Public Services and AWS followed to develop the Infosys Health Insights Platform and its key components.
Realizing the Importance of Healthcare Data
Healthcare data started to evolve when organizations realized the need to manage the information of their members as assets and not just data.
The concept of a “system of records” like Medicaid Management Information System (MMIS) and Electronic Medical Records (EMR) was introduced to fill this need.
Organizations also quickly recognized the value of facilitating seamless interactions and orchestration of these records with other touch points, and developed “systems of engagement.”
In the mainframe era, the two systems were tightly coupled. The internet era de-coupled them. Big data and data lakes opened up new possibilities to ingest and store multi-format data from multiple sources.
While this helped gather enormous amounts of data, it did not serve the purpose of meaningfully associating that data to derive intelligence. Most of the data lakes turned into data swamps.
The Struggle to Manage Evolving Healthcare Data
Cloud, open source, and data harmonization technologies gave birth to “systems of information.” At the same time, development of advanced data science techniques such as artificial intelligence (AI) and machine learning (ML) introduced “systems of intelligence to manage the issues related to data swamps, and accelerate growth and innovation.
However, cultivating data science innovation within an organization isn’t easy. Government healthcare leaders frequently complain about their inability to generate the right intelligence, deliver a unified data experience across the organization (perhaps prompting a need for “systems of experience”?) and advance mission outcomes with the data they have.
This is due to one or more of the following challenges:
- Data jam: Government healthcare agencies find it difficult to address how slowly data moves across the organization. They have complex governance processes and lack the right infrastructure to enable data collaboration across departments and agencies. Most of their analytics projects operate in silos, and intelligence gained is inadequate for effective decision making.
- Inadequate analytics maturity: Most of the agencies are still in the infancy of their analytics maturity. Analysis is primarily spreadsheet-driven and without self-service capabilities. The few agencies that do use advanced techniques are able to analyze only a fraction of the data they have.
- Data advocacy and experience: Government healthcare organizations are doing an outstanding job capturing and aggregating terabytes of data. This includes data from the Health Data Initiative, population health data from claims, electronic medical records and prescriptions, labs, Social Determinants of Health, caseworker notes, etc. However, they struggle to turn it into actionable insights and share it in meaningful ways. The absence of a single source of truth hinders their ability to achieve enterprise-wide, uniform, data-driven intelligence.
- Data literacy: Challenges in attracting and retaining high-end data science skills (data scientists) makes it difficult for agencies to take advantage of advanced data science techniques, and to run analytics projects at scale.
- Data optimization: The majority of government data is unstructured and handwritten. This complicates the analysis and consistency of reporting. It’s extremely difficult to semantically combine and harmonize different datasets to create an integrated longitudinal record and generate actionable recommendations.
- Data quality, security, and privacy: There is very little data standardization at government healthcare agencies. Citizen records are often duplicated, and the use of primitive technologies risks exposure of personally identifiable information (PII) and protected health information (PHI), as well as data loss and leakages.
Turning Evolving Healthcare Data into Insights and Actions
Infosys Public Services and AWS decided to solve these problems by building a cloud-native, end-to-end, self-service driven, advanced data and analytics platform.
The Infosys Health Insights Platform helps government healthcare agencies graduate from a legacy, isolated data set to a connected and modernized data ecosystem. By using the platform, agencies can navigate from data to insights to next best actions, driven by automation, and with minimal dependency on data science skill sets.
Infosys Health Insights Platform enables easy access to advanced data science and automated ML techniques to deliver AI-driven next best actions to all the stakeholders in the care continuum. These actions enable stakeholders to define and deliver the right interventions to the right people at the right time, leading to improved care outcomes.
The platform leverages multiple AWS components such as Amazon Simple Storage Service (Amazon S3), AWS Lambda, Amazon EMR, Amazon Athena, Amazon Simple Query Service (SQS), Amazon Simple Notification Service (SNS), Amazon Elastic Compute Cloud (Amazon EC2), and Amazon Redshift.
The platform is capable of scaling up to manage large volumes of polymorphic data end-to-end in a driverless manner (from ingestion to normalization).
Infosys Health Insights Platform is constructed using a plug-and-play architecture guided by the following design principles:
- Cloud-native from inception to development to productization.
- Support for polymorphic data and polyglot persistence and readily capable of ingesting any type of data, in any format, from any source. The data processing logic has been externalized using a JSON-based configurator.
- API-first approach for easy data consumption and cross-agency collaboration.
- Driverless data harmonization for faster time to market and data readiness.
- Built-in observability for continuous tracking and monitoring of the data pipeline.
- Built-in mechanism to self-heal any data processing failures.
- Security and compliance for data both at rest and in motion.
- Automated data wrangling, machine learning, and predictive model management that scales up citizen data scientist skills within the organization.
- Purpose-built data visualization driven by natural language processing (NLP) through native integration to any data source, including NoSQL, SQL, RDBMS, etc. By native integration, we mean no extract, transfer, load (ETL) operations and no Open Database Connectivity (ODBC) drivers.
- “Next Best Actions” orientation to enable digital, structured decision-making (decisioning).
- Scalability to accommodate emerging AI solutions like conversational AI, digital avatar, augmented reality/virtual reality (AR/VR), and blockchain.
It’s important to flesh out how the platform’s core technology would be designed to help government healthcare agencies become more data savvy in this era of rapid digitization. The following diagram depicts the platform’s overall architecture, with a data pipeline that extends across the continuum of analytics system maturity on AWS.
Figure 1 – Infosys Health Insights Platform architecture.
Infosys Health Insights Platform comes bundled with an integrated NoSQL database (Couchbase) and Amazon S3 as the data persistence engine and object store. The platform has been architected to support polyglot persistence and to process data (from ingestion to insights) in a driverless fashion.
How it Works
The data lands in the Amazon S3 storage bucket, after which the pre-processing workflow gets triggered to parse each data feed. An observatory layer, which follows an “event sourcing” architectural pattern, is created to provide a real-time view of the status of each data feed. This transparency makes it easy for users to understand where their data is in the processing cycle.
The platform has a built-in mechanism for polymorphic data ingestion and doesn’t require any separate ETL tools for data integration. It natively automates most of the processes related to data capture, integration, data set enhancement, and data warehousing.
It also combines industry-specific templates and best practices for rapid integration of data from source systems, and brings best-of-breed components to create master data systems. This enables purpose driven business intelligence and advanced analytics.
At the end of the process, the platform ensures all of the data types are harmonized, consolidated using a canonical model based on Fast Healthcare Interoperability Resources (FHIR), and stored securely in a persistence layer for consumption. The golden record set/consolidated data can be accessed through APIs to populate reports and dashboards, run descriptive analytics, send notifications, and develop predictive models.
The platform’s unified business intelligence (BI) capabilities through Knowi are powered by NLP that enables non-technical users to query the platform for answers, and quickly create purpose-built dashboards. Through elastic search within the user interface, the platform provides flexibility to visualize data in a manner that’s easy to use and enables data-driven decisions.
The platform’s predictive analytics and advanced data science components automate the process of model building, model selection, model deployment, and model management.
The automated ML component offers the following key capabilities:
- Optimized feature engineering.
- Comprehensive library of model test algorithms.
- Automated time series analysis.
- Automated model validation and custom model creation.
- Monitoring of models in production.
- Support for Python extensions.
This robust architecture allows agencies to use the platform as a one-stop shop to build a next-generation data-driven organization and improve outcomes across areas like opioid addiction, mental health, behavioral health, payment integrity, population health management, epidemic management, child welfare, and more.
Citizens are demanding more from their agencies when it comes to health and human services programs. Agencies are rethinking how they can deliver more personalized care to citizens, proactively manage their risk conditions, and monitor their wellness to improve health outcomes.
Infosys Public Services (IPS) and AWS are collaborating to help agencies rethink service delivery. Our next-generation data science platform enables government healthcare agencies to develop the integrated data analytics capabilities needed to improve outcomes. It will lower cost, strengthen population health interventions, and increase stakeholder satisfaction while building an enterprise-level advanced data science foundation.
Infosys Health Insights Platform enables agencies to disrupt (in a positive manner) the delivery of health and human services.
Infosys – APN Partner Spotlight
Infosys is an APN Premier Consulting Partner and Managed Service Provider. Infosys helps enterprises transform through strategic consulting, operational leadership, and co-creation of breakthrough solutions in mobility, sustainability, big data, and cloud computing.
*Already worked with Infosys? Rate this Partner
*To review an APN Partner, you must be an AWS customer that has worked with them directly on a project.