NLM logo

National Information Center on Health Services Research and Health Care Technology (NICHSR)

HSRProj (Health Services Research Projects in Progress)

Information about ongoing health services research and public health projects


Semiparametric causal inference methods for adaptive statistical learning in trauma patient-centered outcomes research
Investigator (PI): Hubbard, Alan
Performing Organization (PO): (Current): University of California, Berkeley, School of Public Health, Division of Epidemiology and Biostatistics / (510) 642-3241
Supporting Agency (SA): Patient-Centered Outcomes Research Institute (PCORI)
Initial Year: 2014
Final Year: 2018
Record Source/Award ID: PCORI/ME-1306-02735
Funding: Total Award Amount: $990,418
Award Type: Contract
Award Information: PCORI: More information and project results (when completed)
Abstract: The revolution in digitizing medical data, along with high throughput biology has resulted in unprecedented numbers of potential prognostic variables. While the medical community is increasingly awash in a sea of highly dimensional and more frequent data collection, this has not yet translated into better prediction or improved outcomes. Our group has led efforts for methods for real-time collection of complex multivariate data of severely injured trauma patients and coupled this with advanced statistical methodology to use this data to define the patient's physiologic "state" space. Our preliminary work has suggested the potential that such high-dimensional data can be used for both continually updated accurate prognostics, as well as help to provide real-time decision support to guide treatment. Specifically, the goal is to harness this information to determine in a dynamic way which treatments are most likely to be beneficial to specific patients. Simultaneously with our preliminary work in the ICU, we have also made significant progress in statistical methodology at UC Berkeley Biostatistics in causal inference and machine learning, which can provide parameters targeted towards specific goals of patient-centered research and the means to estimate these with very high-dimensional data. Our proposal here thereby provides a unique marriage of new opportunities in patient data collection and advances in statistical theory and computation. This proposal combines the expertise of physicians/surgeons directly involved in patient care, along with computational biostatistics to develop methods for targeting patient-centered parameter estimation. Our ultimate goal is to leverage new advances in statistical theory for creating real-time decision tools recalibrated for individual patients based on their characteristics, which can provide continuously updated prognostic information along with estimated impacts of treatment decisions, thereby providing decision support to improve care delivery and outcome. The data used to construct and calibrate the analytical tools are from prospective observational studies of trauma patients in the U.S. and cardiac surgery patients in France. The data include complex physiological measurements and biomarkers. Statistical methodology is based on utilizing recent progress in statistical computing and methods developed to predict the impact of treatment for individual patients. By harnessing the recently available ICU data stream, combined with progress in statistical computing, we plan to develop tools for improving patient outcomes in stressful, dynamic clinical environments.
MeSH Terms:
  • Artificial Intelligence
  • Biomarkers
  • Biostatistics
  • Cardiac Surgical Procedures
  • Decision Support Techniques
  • France
  • Humans
  • Learning
  • Models, Statistical
  • Outcome Assessment, Health Care
  • * Patient-Centered Care
  • Prognosis
  • Prospective Studies
  • * Statistics as Topic
  • United States
  • Wounds and Injuries /*therapy
Country: United States
State: California
Zip Code: 94720
UI: 20143598
Project Status: Completed
Record History: ('2018: Project extended from January 2018 to August 2018. 2017: Project extended to 2018.',)