Episode: AI Is a How, Not a Why, Rethinking Product Strategy for Health Plans

AI is a tool for reaching a goal, and health plan leaders get the most from it by starting with the problem they need to solve. Chris Lance, Chief Product Officer at Quest Analytics®, and Adam Sullivan, an independent consultant who previously led data science work at Change Healthcare and Labcorp, explain how to tell when AI is the right tool for health plan provider network management. 

The conversation gives you a way to sort those goals by what AI actually changes for the business and to spot where it does not belong. 

What You’ll Learn 

• A practical way to tell whether AI is the right tool for a given problem 

• Where AI has opened up work that used to be out of reach for health plan network teams, and where it creates new problems to manage 

• Why the data underneath an AI feature matters more than the model, and what that means for build versus buy 

• How to measure whether an AI feature is working six months after launch 

When Health Plans Should Use AI 

Start with what matters to the business and what it is struggling with, then apply AI to that problem. Chris describes three roles AI can play, which are creating new problems to solve, solving problems that were out of reach, and improving jobs already being done. A rules-based method that returns the exact answer is the better fit for decisions such as care coverage and reimbursement. 

Why Data Matters More Than the AI Model 

Flawed data produces wrong predictions and decisions that are not good for members. For example, ghost networks show the risk. A network can look adequate on paper while the providers listed in it are not available to members. 

Build Versus Buy for Health Plan AI 

Chris says the owner of the outcome should decide whether to pursue the work at all, and a viable product that already exists is better bought than rebuilt. Adam adds that being able to build something does not mean a team should. He suggests the strongest AI talent belongs on high-value problems rather than on replacing a software subscription. 

How to Measure Whether an AI Feature Is Working 

Set a goal and capture a baseline before launch, then measure progress against both. Adam recommends building monitoring in on day one and tracking the value created alongside token spend. 

Related Resources 

Listen to the previous episode in this series on the data science perspective. Read more on the Quest Analytics blog about ghost networks and the Meet Quinn™ AI assistant.