OBJECTIVES: Artificial intelligence (AI) allows the user to harness machine learning capabilities to discover unbiased multi-dimensional relationships in datasets. Advanced data analysis and visualization, powered by AI, were used to identify relationships and trends in a US price migration database, otherwise obscured by dataset size, multi-dimensionality and the limitation of hypothesis driven data analysis.

METHODS: A proprietary price migration database of oncology products tracking prices from January 2004 to June 2018, developed by GfK’s market access health team, was inputted into IBM Watson Analytics. Data curation, including aggregation methodology was defined for the dataset. Data discovery was performed, whereby multivariate relationships in pharmaceutical pricing data sets were extracted and analyzed in an unbiased high-throughput fashion utilizing natural language questions. Outputs fed analysis by our team of market access experts to evaluate drivers of price changes over time.

RESULTS: Visualizing multi-dimensional relationships and predictive drivers for further analysis provides clues to multi-criteria relationships in oncology product price hikes based on indication, orphan status, route of administration, tumor type, companion diagnostic, FDA approval pathway, compound annual growth rate (CAGR), wholesale acquisition cost (WAC), and frequency of update. Visualization of these trends allow for further sub-analysis of data.

CONCLUSIONS: Data discovery utilizing AI greatly simplifies multi-variate analyses as an input for conventional market access analyses. In addition, by moving beyond traditional two- or three-dimensional plots, AI exceeds conventional methodology for rapid visualization of price migration data subsets.