Supernova Award Category

The Problem

With a growing client base, eviCore wanted to scale its business without slowing down approval cycle times for patients and providers or significantly increasing the workload of nurses and doctors who review cases. The company zeroed in on advanced analytics as a way to support decision-makers and speed the benefits approval process.

Previously, the data processing step alone would take several months and required specialized data science and IT resources to complete. That’s no easy task when you’re the size of eviCore, processing 150,000 cases per day into one’s system: from insurance claims to medical procedures and patient data.

Additionally, data processing differed by business unit which prevented standardized reporting of performance indicators -- both internally and to customers -- as well as the operationalization of analytics throughout the organization.

The Solution

eviCore needed an advanced analytics solution that, with minimal IT involvement, could quickly process large volumes of available data and push actionable insights to the decision-makers reviewing claims.

eviCore selected Alpine Data because it addressed three important aspects of its analytic vision on one platform:

1) A collaborative platform that allowed eviCore to standardize data intake, processing and reporting across business units

2) In-database processing and PMML export to allow for efficient database querying, real-time claims scoring, and operationalized analytics

3) An accessible and intuitive user interface to enable non-data scientists to perform advanced analytics in a way that could be scaled throughout the organization

Matt and his team have completely revamped how the company analyzes its data; allowing them to focus on helping ensure the right patient gets the right clinical procedure at the right time and reduce the level of administrative work required in the process.

The results

eviCore was able to easily and efficiently deploy predictive models in Alpine that leverage patient, procedure and insurance data to speed up the pre-approval of incoming claims. These models enabled eviCore to score claims based on the likelihood of being either approved or denied, and, most importantly, to operationalize these predictions at the point of action.

A key benefit of the Chorus platform is that it helped eviCore close the gap between analytic model
 development and operationalization. Now, the process of preparing and modeling data is centralized on Chorus so that data analyst teams can quickly understand a business unit’s problem, process their data, construct a model, and deploy that model into production in a matter of weeks, without hiring data experts with specialized PhDs. Previously, the data processing step alone would take several months, and required both data science and IT resources to complete. This implementation takes the burden off of IT and through the implementation of one platform, has the potential to disrupt the way businesses approach data science and predictive analytics. 

Metrics

The first phase of eviCore’s benefit management process automation lead to multi-million dollar savings on a $125,000 investment in Alpine licenses -- a 500% return on investment. Cost savings resulted from:

--Completing the data processing step in a matter of weeks, not months

--Building and deploying models without needing to hire specialized PhDs

--Decreasing the demand on clinical decision makers by routing claims with a high likelihood of approval / denial straight to the doctor

--Automatically identifying incomplete claims and returning for completion

--Reducing the amount of time it takes health provider offices to get approvals

Additionally, eviCore is now able to integrate analytics with existing end-user workflows, so that machine learning can drive actual business outcomes, such as improved efficiencies and faster claim approvals. For example, analysts can quickly build scoring models and export them for use without needing to write complex code. They are able to capture input and collaborate with various stakeholders in the analytics process in one centralized location, allowing them to deploy useful analytics faster, with fewer wasted cycles.

The Technology

Alpine Data's Chorus platform

Disruptive Factor

Matt had a vision of transforming his organization into one that could use data science to drive operations -- by automating large parts of the benefit management process and reducing the demand on skilled clinical decision makers. In order to integrate data science at the organizational level, Matt believed he first needed to demystify big data and the tools of data scientists. He saw big data not as something new, but rather as an issue of accessibility and scale. With the right tools, Matt believed someone with a business background and intermediate analytic ability, who fundamentally understood the business problems eviCore needed to solve, could use data science to find solutions in a way that was accessible to other analysts and could be scaled throughout the organization.

To prove this concept, Matt hired a recent Masters in Applied Economics graduate, Daniel Ortiz, and tasked him with defining a use case and solving a business problem using the tools of data science. In a matter of weeks, Daniel was able to use the Alpine platform to automate part of the claims approval process and save eviCore millions in expensive nursing and physician hours. The success of Matt’s proof of concept project convinced the organization that this wasn’t a science project, but something with substantial business value that could be replicated by other business units within the organization.

Shining Moment

At the proof of concept stage, Matt asked new hire Daniel Ortiz to blog about using the Alpine platform on an internal company site. His blog described how experimentation with different operators could be used by non data scientists to solve business problems. The blog was an internal hit and went a long way towards Matt’s goal of demystifying data science within his organization.