Case Study 1: Developed a model to predict patient discontinuation for an oral drug using specialty pharmacy data

Business Problem

Client wanted to improve patient adherence. They wanted to use the rich specialty pharmacy data to provide specific and actionable insight to the field teams to reduce patient discontinuations.

ProcDNA Approach

Prediction Algorithm

The ProcDNA team developed a prediction algorithm to estimate the risk of patient discontinuation based on several factors such as physician and patient profile, physician dose modification behavior, fulfilment trends seen in specialty pharmacy data

Decision Tree

We developed logistic and decision tree regression models using the number of patients discontinued (dependent variable) and different independent variables

Risk Score

Finally, we developed a risk score “1-5” (5 = high risk) for all patients in specialty pharmacy data, shared scores with field teams

Client Impact

Real-time Guidance

Provide real-time guidance to field teams on physicians with high-risk patients for discontinuation


Measurable Success

Reduced patient discontinuation by 14%


Let’s Get To Work