Virta Health transforms Type 2 diabetes care with Google Cloud

The pandemic shined a light on the metabolic disease crisis, especially here in the United States for the millions of Americans with type 2 diabetes. According to the American Diabetes Association, 11.3% of the U.S. population (37M+) has diabetes and an estimated 8.5 million Americans are undiagnosed, leaving them vulnerable to complications such as amputation, blindness, kidney damage, heart disease, and premature death. The financial burden isn’t small either–the total economic burden of type 2 diabetes in the U.S has exceeded $400 billion annually, an average of $13,240 per diagnosed individual. We saw the deadly impact that having type 2 diabetes can have on those that tested positive for COVID-19 as 40% of those that died from the virus had diabetes. And per CDC data, 78% of patients hospitalized with the virus were overweight or obese. 

Virta Health is working to find solutions to fight this epidemic with its clinically-proven treatment to safely and sustainably reverse type 2 diabetes and other chronic metabolic diseases, without the use of medications or surgery. To help patients reverse their condition, Virta provides access to a dedicated virtual care team and structured education that makes it easy to understand how certain foods affect chronic high blood sugar and then make the necessary dietary changes. As these changes take hold, Virta safely de-prescribes medications. By combining innovations in nutritional biochemistry, data science, AI, and digital tools with clinical expertise, Virta brings together a unique blend of technology, physicians, and health coaches via an easy-to-use app that’s powered by Google Cloud.

At Google Cloud, we see our role as helping Virta’s Patient centered Ecosystems in three ways:

1. Bringing flexibility and control to Virta’s infrastructure

Since first launching in 2015, Virta has grown from a small healthcare startup to an established business that has treated tens of thousands of patients with a goal of reversing type 2 diabetes in 100 million people by 2025. As it has grown, so has its infrastructure needs. In 2020, Virta decided it was time to migrate its continuous remote care platform from compliant container orchestration platform Aptible to Google Cloud’s managed Google Kubernetes Engine service. There were several main motivations for this switch from Aptible to GKE:

• Greater Optionality: Flexibility and control within networking, scaling, and tooling decisions. 

• Detailed Permissioning: Fine-grained ability to grant permission and access to various workflows while restricting others, which is crucial as its engineering team grows.

• Managed Tools: Access to industry-leading managed tools such as Cloud Functions, BigQuery, and the Google Cloud Healthcare API.

2. Establishing an Interoperable Data Platform and Information Layer

This layer deals with data capture, curation, management, storage, and interoperability to create a common data set upon which the ecosystem can operate. As Virta moves more workloads into FHIR stores, the Google Cloud Healthcare API provides a functional, compliant, and validated FHIR store out of the box, and was a key motivating factor for the migration.Google Cloud’s Cloud Healthcare API helps establish a solid underpinning  to enable interoperability and unleash Virta’s innovative ways to collate patient data leveraging open standards. The first step in treating and reversing diabetes is to identify the population at risk. By leveraging BigQuery, advanced analytics can be performed and data can be further enriched by combining metabolic markers, nutrition and medication data.

3. Enabling Development of Intelligent Systems using Artificial Intelligence and Machine Learning 

Machine Learning (ML) and Artificial Intelligence (AI) play a key role in systems that enhance treatment workflows for clinicians and the overall care experience for patients.

Providers (i.e. doctors, nurses, and health coaches) are working one-on-one with patients daily. They help educate patients on how to reverse diabetes and then put those learnings into practice everyday. In order to personalize care, providers need to be armed with information in real time. Virta’s providers engage with patients reactively and proactively. Reactive care could be in response to patient requests for support, such as answering patient questions about dietary choices. Examples of proactive care include reaching out after noticing a lack of blood glucose values recorded, celebrating an achievement, or making medication adjustments as dietary changes take effect.

Using numerous Google Cloud products (Kubernetes Engine, BigQuery, Google Cloud Storage, Cloud Build, and more) Virta’s ML engineers have developed AI systems that comb through de-identified data like messages, biomarkers, labs, and health outcomes in order to identify patients that would most benefit from proactive care. These AI systems are part of a pipeline that ultimately alerts clinicians about “patient engagement opportunities” and provides contextual information to guide personalized outreach. The pipeline is being continuously iterated on by a cross-functional team of Virtans including applied AI, clinical operations, design, and product. 

Using ML, Virta can flag which patients are most likely to leave the treatment so providers can intervene early to keep patients engaged. They also use ML for estimating a patient’s HbA1c (a measure of blood sugar) to determine current progress toward reversal. Based on these inputs, along with other information, alerts are surfaced to clinicians who can then decide how to act. This is vital toward keeping Virta’s growing panel of patients safe as they move through the treatment. 

To prioritize engagements, the team takes a hybrid approach that mixes rules and ML techniques, giving higher priority to:

• Engagement opportunities that are most time-sensitive.

• Patients who have not received a message from a coach recently.

• Patients who began treatment more recently to ensure they are off to a good start.

In 2021, the team focused on health coach alerts and ways to increase the percent that led to coach action. They modified engagement opportunity criteria to be more strict where there were too many unuseful alerts, adjusted the prioritization algorithm, and introduced new alerts to fill gaps that had been identified by health coaches. As a result, health coach action increased by about 20%, enabling coaches to be more efficient and effective, which in turn has led to greater patient success.

We’re excited to be such a large part of helping Virta achieve its goal of reserving Type 2 diabetes in millions of patients. In 2022, the company anticipates triple-digit year-over-year customer and patient growth. Increasing patient volume will require Virta to lean even more heavily into ML & AI to keep patients safe, engaged, and successful on the diabetes reversal treatment, while continuing to streamline workflows for clinicians and health coaches.