Prediction of Hospital Readmissions for Precision Medicine


When patients leave the hospital after surgery they expect to be on a path to better health. Unfortunately, up to 15% of patients are unexpectedly readmitted to the hospital within 30 days of discharge. Hospital readmissions are an enormous public health concern and cause excessive strain on healthcare resources. Readmissions for musculoskeletal and heart conditions in particular are both frequent and expensive, costing billions in medical expenditures each year. Orthopaedic and heart surgeries are in the top 20 costliest procedures. In order to reduce excess hospital readmissions, prediction models are needed to identify patients at high risk of readmissions and to intervene in a timely manner. Prior attempts to develop risk calculators have been limited by insufficient data sources or results that could not be tailored to an individual. To date, there has not been any models developed that have the capability to model hospital readmissions for orthopaedic and heart conditions, for all ages and for all payers using big data repository linking clinical and non-clinical variables for precision medicine. The proposed research seeks to develop models that can predict which patients will experience hospital readmissions in an effort to identify targets for intervention. Once completed, this work has the potential to improve health care quality, reduce cost, and improve patients' quality of life.

Student Role

Students will be involved in each phrase of the project from study design, literature review, data management and analysis, to drafting manuscripts and preparation for journal publication and conference presentation. As students develop research skill competency, greater autonomy will be given in completing research tasks. Students also have the opportunity to learn and participate in other ongoing research projects to broaden their knowledge and perspectives.


Students participated in this project will be actively learning the proper designs of scientific research. They will be trained in the ethnical and practical aspects of human subject research and data management. They will learn how to conduct literature review, cite references, use statistical software to perform appropriate data analysis, write abstracts and manuscripts for publication and deliver professional presentations. They will have the opportunities to be co-authors on abstracts and manuscripts. These hands-on activities will increase students' research skills and will be beneficial for students who seek further education in medical school or other graduate-level programs.

Man Hung
Associate Professor

College of Medicine

While a major goal of this project is to develop models that can predict which patients will experience hospital readmissions in an effort to identify targets for intervention, a second and equally important goal is to provide students with rich opportunity to participate in mentored research. To facilitate both the project and the mentoring aims, a regular meeting with all mentored students will be held to provide direction, instructions and guidance regarding the research process, to provide training and to provide skill building opportunities. Such meetings will allow for frequent feedback regarding students' work.

I maintain professional and personal relationships with many of my former undergraduate and graduate students and continue to provide letters of recommendation, mentoring and advice. Cultivating long-term relationships with students provides potential future collaborators as the students move through their academic training and on to graduate school and academic careers. My goal is to form long-term relationships with my students who will become my future colleagues and early stage mentoring is an excellent way to build these relationships.