(US & Canada) | Governance Framework Is the Constitution for Data Democracy — Horizon BCBS of New Jersey Chief Analytics Officer

Vijay Venkatesan, Chief Analytics Officer at Horizon BCBS of New Jersey, speaks with Ron Rogers, Principal Consultant at AHEAD, in a video interview about strategies to overcome data management issues while leveraging AI, data democratization, and having a solid operating model, an enterprise approach to integrating health data, and data use cases to protect patients at risk.

Horizon Blue Cross Blue Shield of New Jersey provides health insurance coverage to more than 3.8 million people throughout all of North, Central, and South Jersey.

The second segment of the interview begins with Venkatesan shedding light on the strategies to address data management issues while leveraging AI. At BCBS, it starts with establishing an AI governance framework that involves privacy, compliance, security, business, and IT stakeholders, collectively framing the guiding principles.

The next piece in this is assessing data completeness, says Venkatesan. He refers to it as a data scorecard and gives an example of the completeness of claims data versus provider data.

Elaborating further, Venkatesan states that if a claim data is 95% complete, the provider data will be 65% complete because most of the provider data is about payment and reimbursement. Thus, there may be data fields that are needed but do not show up, such as the billing address of a provider, their practice location data, and so on.

Consequently, the use cases that pass the evaluation criteria of data completeness are taken into consideration, says Venkatesan. In that, if the data indicates bias, then the same is assessed responsibly. The framework also examines the need for creating or buying LLMs, its pitfalls, and whether or not to internalize data.

Venkatesan opines that the framework starts with governance, then gets to the use cases with a high degree of data completeness, adding the responsibility component, and then operationalization comes into play. The critical aspect to focus on is whether the operations sector is prepared to use the insights and create meaningful actions, he notes.

Delving deeper into the data completeness piece, Venkatesan mentions having a data quality framework to measure data quality. Further, he discusses doing an overall reasonability scan that examines what purpose the data serves. For instance, assessing if a data source is good for population health.

In addition, to make data more usable, Venkatesan uses traditional predictive models to build member-level attributes. He refers to creating a “feature store” around a member that could enhance the eligibility file and data quality.

On the responsible use side, Venkatesan works with the BCBS association to look into the established guiding principles and look at other industry benchmarks to bring in perspectives. According to him, it boils down to understanding the demographic while using data and examining external and internal data resources to understand where bias lies.

When asked if the healthcare industry could use data-as-a-service, Venkatesan discusses data democratization. The key pillar in democratizing data is the operating model for doing data and analytics. Considering the evolution of the operating model, he predicts that the industry will have more of a hub-and-spoke operating model.

In that hub-and-spoke model, the industry would foster a notion of data democratization, wherein many common rules would be homogenized or centralized. Again, there would be “spoke” functions and accountability to have democratization at scale.

Adding on, Venkatesan analogizes, stating that there is no democracy without a constitution. In this case, the governance framework is the constitution that enables data democracy. While the hub should maintain, manage, distribute, and educate, the spoke should be able to adhere to it within that framework while retaining its voice. That is the model the industry is trying to emulate and is in its early stages, he adds.

Moving forward, Venkatesan states that he is trying to create a model that can help build the best data distribution supply chain without creating chaos, for instance. While some say he’s building a medallion architecture, and others call it data fabric, he intentionally veers away from the buzzwords.

Highlighting a good enterprise approach for combining multiple types of health information data, Venkatesan mentions the need to assess organizational information maturity levels. For some organizations, the regulatory issue drives the impetus for data standardization, which forms one kind of data integration framework.

Whereas, if it is a prolific data organization that wants data supply to be a constant and evolving iterative framework, it can come up with a different integration model. According to Venkatesan, if an organization owns a data catalog and works with data producers to create an integration framework, it is on the right track.

Furthermore, Venkatesan considers catalog building a central function. This process involves establishing a central catalog within organizations, akin to an initial version or "iterative 1.0," which serves as a framework for integrating diverse data.

It's likened to deciding whether to develop a data product architecture, with options ranging from custom to standard products, shares Venkatesan. The catalog, analogous to an iTunes store, represents this data product.

Once established, it helps determine if custom data products are necessary. This approach simplifies organizing efforts within organizations, facilitates the categorization of data producers, and incentivizes them to contribute to a central repository.

Commenting on data use cases that actively protect patients at risk, Venkatesan states that as part of the value-based contracts, organizations try to look into care in general. By identifying patient risk, gaps in care, and so on, organizations create a bidirectional data exchange that feeds the data back into their electronic health records in the provider system.

Also, the organization looks into social health determinants and merges that information to figure out factors that the provider must consider to get patients the required care.

In conclusion, Venkatesan states that organizations must be able to impact both the member side and work with providers on the patient side. If the goal is to reduce cost, improve access, and improve overall health quality, it must be a continual dialogue.

CDO Magazine appreciates Vijay Venkatesan for sharing his insights with our global community.

Also Read
(US & Canada) VIDEO | GenAI Is a Tool in a Toolbox for Solving Business Problems — Horizon BCBS of New Jersey Chief Analytics Officer
Vijay Venkatesan, Horizon Blue Cross Blue Shield of New Jersey Chief Analytics Officer

Executive Interviews

No stories found.
CDO Magazine