Leveraging Machine Learning for Hospital Segmentation based on potential caseload: A Case Study Approach

This case study focuses on using machine learning to segment hospitals into potential clusters based on their number of patients treated for a certain disease. The importance of this task is relevant for pharmaceutical companies for optimizing salesforce allocation. Pharmaceutical companies employ sales representatives to influence physicians in prescribing their drugs, making thus crucial to prioritize hospitals that have higher caseload potential for minimizing travel expenses and operational costs. However, obtaining the necessary healthcare data for informed decision-making can be challenging and costly and consequently pharmaceutical companies rely on external companies like IQVIA, a leader company in the life-science information technology sector. The objective of this study is therefore to develop a machine learning pipeline for IQVIA that estimates patient volumes of a certain disease in hospitals that are not part of the universe mapped by the market research team. Given the presence of a labeled dataset from which a machine learning model can learn, the effectiveness of supervised machine learning models in accurately estimating patient volumes will be evaluated. Moreover, the proposed pipeline will be tested in different datasets concerning different diseases since the ultimate goal of this study is to assess the potential for automating the hospital segmentation process, as the underlying variables required for model construction remain consistent across diseases. A successful development of a scalable machine learning pipeline could significantly speed up the analysis required in these types of projects and be beneficial for the IQVIA company or similar companies. This research encompasses a comprehensive literature review, data collection and preprocessing, experimentation and model development, testing on different datasets, analysis of results, and exploration of potential variations and extensions. By critically evaluating the outcomes and addressing challenges, this study provides valuable insights into the applicability of machine learning in hospital segmentation for the pharmaceutical industry.