Our Work - Blue Cross Blue Shield Provider

Churn Prediction Models Improve Health Insurance Customer Retention.

 
People Running

Roles

  • Project Leadership
  • Data Engineers
  • Data Scientists

Deliverables

  • Exploratory Analysis
  • Reproducibility Report
  • General Information Plots
  • Correlation Matrices
  • Preprocessed Data Sets
  • Classification Model
  • Results Documentation
People Running
OVERVIEW

A Success Story

In the health insurance industry it can be extremely challenging to find useful indicators of unhappy customers outside of direct feedback from customer service calls or surveys. Yet, keeping customer churn as low as possible is extremely important, since the cost of acquiring new customers is steeper than retaining existing ones. Any improvement in customer churn has a big impact in revenue.

A Blue Cross Blue Shield provider in Puerto Rico turned to Wovenware to create a proof of concept for predicting churn among Platinum Advantage members with claims and demographic data. After completing three Innovation Sprints, Wovenware was able to create an AI-driven predictive model with 93% accuracy.

THE CHALLENGES

What We Faced

  • Identify members with high probability of changing providers prior to the open enrollment period based on a limited dataset of claims and demographic data.
  • Obtain optimal classification results when working with imbalanced datasets resulting from a much lower number of disaffiliations compared to existing customers.
  • Employ business acumen to balance a dataset by engineering new features without changing the real-life outcomes described by the original dataset.

Working on challenges

 

OUR APPROACH

How We Helped

Wovenware used its Innovation Sprint methodology to examine the quality of the data and determine the feasibility of creating a churn prediction model.

  • Preprocessed data to consolidate claims and performed an exploratory data analysis.
  • Initially did not find any meaningful features to train a deep learning model given that all demographic and claims data followed the same distribution.
  • Engineered new features to balance the dataset based on knowledge about the client’s data.
  • Employed a deep learning model consisting of three fully connected layers, a single neuron at the output, and the sigmoid activation function.
  • Optimized a binary cross entropy loss using a sigmoid output.
  • Used a holdout set to validate the model and avoid statistical bias.

Circles textureCircles texture

AI Model

Technologies

  • Keras

    Keras

  • Pandas

    Pandas

  • NumPY

    NumPY

  • Scikit

    Scikit-learn

  • Jupyter

    Jupiter Notebooks

RESULTS

Client success

Wovenware validated the potential impact of a custom model for predicting patient turnover with a proof of concept based on a subset of sample data that had outstanding accuracy. The custom predictive analytics model identifies members who are most likely to churn, giving the BCBS provider critical information to target specific populations in customer retention strategies. Artificial Intelligence is playing a key role enabling the organization to transform the patient experience through proactive remediation. 

  • Recall: 93%
  • Precision: 78.23%

 

Digital notepad