Enterprise AI for Managed Healthcare Organizations

Challenge

The company had seen a dramatic increase in the prevalence of opioid dependence among members, leading to increased costs associated with treatment and lost productivity. To reduce costs and provide targeted care to members, the company’s internal data science team had utilized claims data to design and develop a classification machine learning model that can identify risk of opioid dependence for each member. The classifier was developed as part of an on-premise solution on an internal cloud and leveraged an existing internal data lake. However, the technical hurdles to develop and operationalize large-scale machine learning applications presented multiple challenges:

Approach

To address the challenges associated with the enterprise-wide application of the company’s risk classification tool to millions of members, the C3 AI team designed and developed the C3 AI Patient Risk application on the C3 AI Suite. Over four weeks, the C3 AI team developed the enterprise level application for classifying patients based on their risk levels for developing long-term opioid dependence, enabling the company to customize preventative and targeted treatments to patients at the highest risk.

About the Company

  • Top five managed healthcare organization in the US
  • Operating across over 150 countries
  • More than $100 billion in annual revenue in 2019
  • More than 120 million medical members globally and 50 million in the US
  • More than 150,000 employees

Project Objectives

  • Integrate and unify data from six disparate sources including membership, pharmacy claims, coverage, facility claims, physician claims, and the company’s existing data lake.
  • Implement and augment the company’s existing machine learning classifier on the C3 AI Suite to enable deployment at scale.
  • Develop a patient risk application to facilitate communication and delivery of information on at-risk members and member groups.
  • Develop microservices such as API’s to expose AI insights and unified analytics to end users such as providers, claims managers, and member service managers.

Results

$700M
in potential economic value from addiction prevention
80%
accuracy in predicting patient risk of developing opioid dependence
40x
improvement in developer productivity
5x
reduction in machine learning algorithm lines of code required

Solution Architecture

C3 AI Suite

Want more information?

Contact us