Drivers of prescriptions

Objective

Understand the most important factors driving drug choices at dierent stages of a patient journey, basing which create favorable patient profiles for NTM and Switch targeting

Approach

  • Data sources: Integrated Claims and EMR data
  • Overview of approach:

  • For Model Building, D cube employed a combination of:

    Strong Predictive Techniques, that leverages machine learning concepts (Random Forest) to get best-in-class models

    Descriptive Techniques, like decision trees to enable attributes classification and help business in deriving marketing strategies from a tree based output

Results

  • NTM Analysis (1st Line): Overall Accuracy of the model : (83%)
  • Key Insights:New-to-market prescription decisions are driven by prescriber’s specialty, risk stratification, biomarker results and comorbidities & concomitances.
  • Impact: Estimated a potential lift of 7% in NTM market share
  • Switch Analysis (2nd Line): Overall Accuracy of the model : (75%)
  • Key Insights: Unlike NTM decisions, which were heavily dependent on disease severity & provider biases, Switching decisions are more strongly driven by patient choices & market access
  • Impact: Estimated a potential lift of 13% in 2nd Line market share