OM Job Talk - Ujjal Mukherjee , PhD Candidate from Carlson School of Management, University of Minnesota will present a talk in the Operations Management Area
Research Seminars
Academic Areas Operations Management
Ujjal Kumar Mukherjee , PhD Candidate, Carlson School of Management, University of Minnesota
January 5, 2015
| 8:30 AM - 10:00 AM | Monday
AC-2 Mini Lecture Theatre, ISB Campus, Gachibowli, Hyderabad, India, 500 111
Open to Public
Topic: “Predicting Failures of High Tech Innovations-in-Use: Application of Predictive Analytics to Big Data on Market Failures of Medical Devices”
Abstract: Most high tech firms today realize that failure of technological innovations while in use in the marketplace, manifested in the form of product recalls, cannot be completely eliminated. Hence, it is of critical importance that early signals of such failures are timely detected. The central question that serves as the motivation for this paper is: Can user-level feedback related to episodic adverse events in the market predict product-level innovation failure risk? The theoretical lenses we have used to frame the research questions into testable hypotheses are signal detection and system neglect. Medical device industry is the empirical setting for this study. The primary dataset we are using for empirical analysis is the Food and Drug Administration’s (FDA) “Manufacturer and User facility Device Experience (MAUDE)” dataset. The MAUDE database represents “big data” generated through reports of adverse incidents involving the usage of medical devices. Using this “big” dataset, we develop a predictive model that can ex ante estimate the hazard or risk of innovation failure. The contributions of this paper are the following: First, we demonstrate that it is possible to predict innovation failure risk with reasonable accuracy and consistency from market data. Second, we demonstrate that it is possible to improve precision of prediction by incorporating knowledge of explanatory variables related to the products, firms and industry. Third, we demonstrate that systematic judgment bias exists in the innovation failure detection process. We derive conditions under which firms exhibit over-reaction bias and under-reaction bias to field failure. Finally, this paper contributes toward extending the boundaries of empirical methodologies in conducting operations and supply chain management research from explanatory regression based models to advanced predictive analytics based models developed from “big” unstructured datasets.
Abstract: Most high tech firms today realize that failure of technological innovations while in use in the marketplace, manifested in the form of product recalls, cannot be completely eliminated. Hence, it is of critical importance that early signals of such failures are timely detected. The central question that serves as the motivation for this paper is: Can user-level feedback related to episodic adverse events in the market predict product-level innovation failure risk? The theoretical lenses we have used to frame the research questions into testable hypotheses are signal detection and system neglect. Medical device industry is the empirical setting for this study. The primary dataset we are using for empirical analysis is the Food and Drug Administration’s (FDA) “Manufacturer and User facility Device Experience (MAUDE)” dataset. The MAUDE database represents “big data” generated through reports of adverse incidents involving the usage of medical devices. Using this “big” dataset, we develop a predictive model that can ex ante estimate the hazard or risk of innovation failure. The contributions of this paper are the following: First, we demonstrate that it is possible to predict innovation failure risk with reasonable accuracy and consistency from market data. Second, we demonstrate that it is possible to improve precision of prediction by incorporating knowledge of explanatory variables related to the products, firms and industry. Third, we demonstrate that systematic judgment bias exists in the innovation failure detection process. We derive conditions under which firms exhibit over-reaction bias and under-reaction bias to field failure. Finally, this paper contributes toward extending the boundaries of empirical methodologies in conducting operations and supply chain management research from explanatory regression based models to advanced predictive analytics based models developed from “big” unstructured datasets.
