Abstract  

Manipal Hospitals, which was started in 1953, had the advantage of being the “oldest” healthcare group in the country.  In six decades, the group could establish the reputation for being ethical and patient friendly. In 2017, Manipal Hospitals catered to around 2 million customers from India and overseas every year through their tertiary and secondary care facilities. In 2017, MHE managed an aggregate of 5,200 plus beds among 16 hospitals, over 13 locations across 6 states in India and one hospital in Klang, Malaysia. The Group's acute care flagship quaternary care facility located in the heart of Bangalore, India's IT capital was set up in 1991. The 680-bed Manipal Hospital at HAL Airport Road provided care in over 60 specialties under one roof.

Ajay Bakshi, MD and CEO of Manipal Hospitals strongly believed that the word-of-mouth (WOM) is much stronger than any other type of promotion and thus it is important for MHE to keep customers informed about the improvements. With the improved system for feedback collection, he was confident of moving towards a more tangible outcome from feedback collection. Collecting Net Promoters Score (NPS) and tracking the trend of NPS was an integral part of patient care at MHE. He also believed that closing the loop is a central theme of the Net Promoter Score and thus NPS should be pivotal to understanding the deficiencies in the system and improving it.

Ajay believed that NPS score itself is just the tip of the iceberg. The real value was provided by understanding what leads to the NPS score, especially the causes of detractors and promoters and asking follow-up questions on the reason for the score. It provided a gold mine of information.

Learning Objective  

The case is useful for demonstrating application of machine learning algorithms to improve customer experience.  The primary objective of the case is to demonstrate how multi-class classification models can be used to analyze data on net promotor score for improving the functions of various departments within an organization and effectively engaging with customers/patients. Other learning objectives include the following:

Demonstrate the application of random forest and adaptive-boosting in solving multi-class classification problems.

1. Demonstrate the superiority of ensemble methods in solving multi-class classification problems.
2. Understand sensitivity and specificity in context of a multi-class classification problem.
3. Learning different types of contrast coding to treat categorical variables.
4. Understand deployment strategies based on machine learning algorithms.