Luma Health Leverages Sisense and Amazon Redshift to Address COVID-19
Luma Health is a digital health company focused on improving patient access and engagement. Using Luma Health’s Total Patient Engagement™ platform, providers and their staff are able to communicate with patients in ways that reduce the traditional friction around scheduling while providing effective modern, mobile solutions for everything from patient intake forms to collecting feedback following a visit. Beyond a much improved patient experience, providers can also leverage metrics collected by the platform to identify best practices for optimizing processes, leading participating providers to outcomes like a 79% reduction in no-shows and an average NPS score of 89.
Long-time Sisense customers, Aditya Bansod, Co-Founder and CTO and Humair Burney, Director of Business Operations and Strategy, have always been keenly aware of the importance of data to their product development and operational workflows. This meant that when the COVID-19 crisis hit, Luma Health’s data science teams were equipped with the analytics capabilities they needed to manage through the crisis. Teams were able to identify patterns in patient-provider communications, rapidly create messaging for providers that addressed the unique challenges presented by COVID-19, and effectively communicated screening solutions to patients, all without providers paying Luma Health an additional dollar.
The challenge: Leveraging AI to address COVID-19
As a healthcare startup focused on using technology to drive better patient experiences and smarter provider interactions, Luma Health has been hyper focused on the value provided by applying advanced statistics and analytical techniques to answer everyday questions. Early success includes Luma Health’s Business Operations team’s analysis of message data via word clouds. This allowed them to listen better, bringing new behaviors to light and empowering the product team to make data informed improvements.
This philosophy and supporting infrastructure meant that they were prepared to respond when COVID-19 began showing up in medical discussions in early 2020. Producing unparalleled challenges for the patient care ecosystem, with COVID-19 Luma Health saw a major opportunity to use machine learning and AI to address some of the critical, unprecedented problems their provider community was reporting, such as how to ensure patients safely remained in their cars at their clinic’s parking lot until their appointment time.
Upgrading urgent provider-patient communications
One of the first challenges Luma Health identified was headaches patients were experiencing when trying to make appointments with specialists or other referred doctors. Many providers still have their staff do scheduling and handle patient calls manually. These tasks suddenly became almost impossible because the extreme health concerns surrounding COVID-19 produced a surge in unidentified callers and robocalls, leading to scheduling failures and frequent cancellations without notice.
These issues especially impacted seniors as they tend to be big consumers of healthcare and they are among the most vulnerable groups at risk for contracting COVID-19. With Luma Health being based out of San Francisco — the first major U.S. city to declare a shutdown due to COVID-19 — the team also had a front row seat to early forecasts suggesting that the health impact would not be short-lived or limited to a discrete geography. With all signs pointing to an impending national crisis, the Luma Health team believed that their community of providers and patients were about to face significant, unprecedented issues that would require new, highly creative technology solutions.
The Business Operations leads at Luma Health quickly convened a special team to figure out what COVID-19 might mean for patient engagement and what Luma Health could do to best respond, leveraging their unique data-driven offerings and their proximity to providers on the frontlines. Luma Health’s leadership team approached the challenge with a belief that the crisis would affect all areas of healthcare, starting with issues as simple as many patients not even knowing if their provider’s office was open for their scheduled visit.
Text analysis with Sisense
This is where Sisense provided insights: Luma Health was able to reexamine its product metrics across the board, to identify opportunities to use technological innovations to provide more support for providers. Luma Health tracks providers’ interactions with over 10 million patients. With Sisense, the team was able to efficiently uncover important actionable patterns from their data on scheduling, medical records management, and patient behavior.
In essence, Luma Health used Sisense to prove out some initial hypotheses. For example, using Sisense’s built-in text analysis capabilities, the Luma Health team noticed that messages containing words like “COVID”, “fever”, “virus” and a similar cohort were skyrocketing. Providers and patients were using Luma Health’s platform from the beginning to communicate and try to make smart health decisions about the emerging crisis so it only made sense that “COVID” was showing up with alarming frequency. Adding on the observed skyrocketing cancellations of appointments across different states in the US and across care specialty lines Luma Health could identify that a core focus on minimizing disruptions to care involved digital communication and tailor those solutions to each segment appropriately.
This insight about the growing mention of “COVID” led Luma Health to release a free broadcast messaging system to assist with COVID-19-related patient-provider communications. Believing that everyone must work together to help during such moments of uncertainty, Luma Health made the new system available free of charge to all people who needed to use the platform, whether they were customers or not.
Expanding the solution scale
Following these initial advances, Luma Health went into “wartime mode.”
With Sisense, data scientists are equipped to quickly build, collaborate and deploy complex and timely data outputs supporting any ad-hoc queries of the data. Using the platform to analyze their data, the Luma Health team discovered insights in the data suggesting that the best way for providers to reach their patients was by SMS text messaging. Among the findings, the data showed that SMS was a much more effective way to reach patients with details about their doctor appointments than robocall voice messages. Luma Health shared these findings with providers, with the hope that they might be able to apply the additional insights to their evolving engagement strategies for patients who had serious concerns about contracting, preventing, and avoiding the spread of COVID.
Looking at the data they were able to analyze, Luma Health also concluded that patients’ responses to messages sent could be used to trigger automatic schedule changes in providers’ systems. This made the whole system a lot more efficient, freeing providers’ staff to do more than just manage manual communications and other tasks related to rescheduling.
End-to-end cloud solution: Combining Amazon Redshift with Sisense
Luma Health was able to effectively roll out their new offerings because they had all of their operational data stored in Amazon Redshift. This meant that a variety of team members, people beyond engineering, had access to the data they needed to be able to perform everything from simple ad-hoc analysis to complex, statistics-based cohort analysis.
The result of pairing Sisense with Amazon Redshift was greater flexibility and speed for hypothesis testing.
Humair Burney, Director of Business Operations and Strategy
Says Humair, “Given the breadth of capabilities of the Luma Health platform across all the points in a patient’s typical care journey and the unprecedented impact of the COVID-19 pandemic on our customer’s operations, we had no shortage of hypotheses to test. When answers are needed and every hour counts, it matters that data is centralized and queries run fast. Amazon Redshift not only provides more flexibility but also runs faster than PostgreSQL, which we used before our analytics needs warranted an upgrade in our warehouse solution.”
Luma Health’s core data is kept in MongoDB Atlas, Redis, and RabbitMQ. To run analytical workloads live, the team leveraged AWS Data Migration Services (DMS) to not only load their MongoDB data sets into Redshift but to use AWS DMS’ Change Data Capture to keep the data in Amazon Redshift live. To better respond to COVID-19 challenges, having a nightly or bulk ETL solution was insufficient since Luma Health moved its analysis time horizon from months down to hours, another benefit of using DMS to keep pipelines live and visible on Amazon Redshift in particular. In the midst of a pandemic, circumstances change hourly so live data is worth 1,000x more than bulk/batch data. By connecting their Sisense instance directly to Amazon Redshift, the Luma Health team enabled quick access to their ever growing data with a real-time “live” feel using the languages they prefer like SQL, R, and Python all in one collaborative application.
Insights to action
Luma Health has rapidly advanced from startup to industry leader in digital health. And with Sisense as their analytics platform and AWS as their cloud provider, their teams are able to release new, timely offerings in a matter of weeks. Regarding the recent COVID-related offerings, Luma Health released them to their community just three weeks after envisioning them, offerings that are still operating now, three months later.
In times of crisis, people need insight. The teams at Luma Health were able to use data from Sisense to analyze, recognize, and address the unpredictable swings in patient behavior that COVID-19 produced, harnessing insights to create a surge of innovation for good.
Aditya Bansod, Co-Founder and CTO