Indranil Bardhan is Foster Parker Centennial Professor of Management in the McCombs School of Business at the University of Texas at Austin. He is a Distinguished Fellow of the INFORMS Information Systems Society. His research focuses on healthcare analytics and informatics, and economic impacts of information technology, and involves close collaboration with the University of Texas Southwestern Medical Center and the UT Dell Medical School.

Predicting ICU Length of Stay using Deep Learning: An Explainable AI Approach

Abstract: Accurate prediction of patient length of stay (LoS) in ICUs is critical to ensure that clinical resources are allocated efficiently for patient care management. In this research, we propose a novel explainable AI framework that predicts patient LoS in the ICU and provides relevant explanation utilizing a combination of ablation analysis and feature attribution through integrated gradient analysis. Our framework can detect the dynamic patterns of feature importance for LoS prediction and identify key model input features that consistently have a significant impact on model prediction for specific patient cohorts as well as for individual patients. Our results suggest that, in addition to claims-based data, inclusion of selected types of clinical data such as patient vital signs, and to a lesser degree, provider clinical notes, can significantly improve overall model predictive performance.

Dokyun “DK” Lee is a Kelli Questrom Chair Associate Professor of Information Systems and Digital Business Fellow at Questrom School of Business. DK is also a fellow at Computing & Data Science School.

InnoVae: Generative AI for Understanding Patents and Innovation

Abstract: InnoVAE utilizes recent advances in generative AI to model patents using both structured and text data. By training the generative AI to represent patents in an interpretable vector space in which dimensions are rendered to be interpretable factors of innovation, InnoVAE enables meaningful comparison between patents and firms that own these patents. Findings illustrate the potential of using generative methods on unstructured data to guide managerial decision-making.

Dr. Ravi Bapna is the Curtis L. Carlson Chair in Business Analytics and Information Systems, Associate Dean for Executive Education and the Academic Director of the Carlson Analytics Lab and the Analytics for Good Institute at the University of Minnesota’s Carlson School of Management. Bapna's expertise lies in helping companies leverage data-science, machine learning, AI and business analytics for competitive advantage.

Social Learning in Prosumption: Evidence from a Randomized Field Experiment

Abstract: While conventional wisdom advocates for increasing consumer involvement in co-creating, designing and producing products with firms, such prosumption is not without its challenges. In our field site, a creative co-creation platform, a significant proportion of consumers start design projects but fail to complete them, and another salient fraction finish designs but do not buy the products they created. We use social learning, the idea that consumers can learn from seeing what their fellow users have done, to reduce frictions in prosumption. Our ideas are tested using large scale in-vivo randomized field experiments and machine learning is used on data from RCTs to get further insights.

Zhe Zhang is an assistant professor of business analytics and economics of digital technology at UC San Diego, in the Rady School of Management. His research focuses is on the societal and broader impacts of new digital technology, including work on the impacts of the sharing economy on business models and urban spending, as well as statistical work on algorithmic fairness.

Growth versus Competition in Growing Digital Markets: Evidence from Amazon Prime's Effects on Competitors

Abstract: My research studies how the e-commerce giant Amazon.com has affected the business of other e-commerce firms in the market. We identify new subscribers of the Amazon Prime service, that provides free faster shipping rates, and examine how adoption of Prime affects their online spending elsewhere in non-Amazon stores. Our research motivation is to study growing digital marketplaces in general, using e-commerce in the last several years as an example.