A Continuous‐Time Markov Chain Model–Based Business Analytics Approach for Estimating Patient Transition States in Online Health Infomediary. (20th January 2020)

• Record Type: Journal Article
• Authors: Lim, Sanghee; Yim, Dobin; Khuntia, Jiban; Tanniru, Mohan
• Is Part Of: Decision sciences. Volume 51:Number 1(2020)
• Page Start: 181
• Page End: 208
• Publication Date: 2020-01-20
• Subjects: Infomediary -- Churn Rate -- Patient Engagement -- Markov Model; Decision making -- Periodicals; Policy sciences -- Periodicals; 658.40305
• DOI: 10.1111/deci.12430
• Languages: English
• ISSNs: 0011-7315

Abstract

ABSTRACT: Online health infomediaries are emerging as a critical element in the healthcare sector to support and influence individuals' health and wellness decisions. The business success and effectiveness of health infomediaries depend on the active and sustained engagement of patients. Although the growth in the number of participants in an infomediary is expected to add value by increasing the diversity of information that is potentially exchanged, the infomediary cannot survive without the sustained engagement of existing users. The challenge is to understand the underlying processes at the operational workflow level of an infomediary that can lead to sustained engagement of patients. For an infomediary to increase engagement, it needs to know not only what motivates participants to join an infomediary but also what keeps them engaged in various stages of participation or transitions. In this study, we employ a Markov Chain modeling approach, along with an analysis of the user activities data, to understand the underlying mechanism of patient engagement along with several transition states in an online health infomediary. We tracked 127, 610 members, with more than 1 million activities involved in an online health infomediary that supports cosmetic and reconstructive surgery patients over one year. Patients' decisions for cosmetic and reconstructive surgery are health and well‐being choices that rely not only on patients' current situation but also on the knowledge and experience of others. This relevance of the health infomediary context is explored in this study. We sampled the activities of 32, 505 active users' activities with data on more than 500, 000 activities. We analyzed the dynamics of user behaviors by modeling longitudinal transition probabilities across different states of participation. Additional analyses and robustness checks, using text‐mined data from the users' activities, are introduced to gain nuanced insights into user engagement. Our study provides several practical implications for the design and management of an online health infomediary.