[[Whitepaper]] Predictive Analytics: A Case Study in Machine-Learning and Claims Databases

In Featured on App, Orphan Drug Congress, Whitepapers by Karina Kusova

Authors: Kvancz DA, Sredzinski MN, Tadlock CG.

Publication: American Journal of Pharmacy Benefits

The study demonstrated the ability of this state-of-the-art predictive analysis to find rare disease patients in a large and complex database, shortening the time to diagnosis and relieving patients of the frustration of going from doctor to doctor without a confirmed diagnosis.


This study focused using the HVH analytics process to find undiagnosed hereditary angioedema (HAE) patients. This disease was selected because it is rare, hard to diagnose, progressive, and is associated with misdiagnoses and underdiagnosis.


The HVH 3-stage process was applied to a claims database in order to:

  • Define patient characteristics (diagnoses, procedures, therapies, and providers) in the database already being treated for HAE
  • Use those characteristics to create a model of patients with HAE
  • Use the model to identify patients with HAE in the database who were not yet diagnosed


This study successfully demonstrated the ability of state-of-the-art predictive analysis to find rare-disease patients in a large and complex database, which can help these patients receive appropriate treatment sooner.

Download the publication and learn more about HVH’s patient finding products and services.