Grant Number:	2R01CA249096-05 Interpret this number
Primary Investigator:	Li, Yi
Organization:	University Of Michigan At Ann Arbor
Project Title:	New Statistical Methods for Modelling Cancer Outcomes
Fiscal Year:	2025

Project Summary/Abstract The proposal addresses two pressing medical problems. First, opioid use is crucial for managing pain in surgical and cancer patients, but over-prescription has led to a significant public health crisis. Opioid overdose deaths have tripled in the past decade, with over half of these fatalities linked to prescription opioids. The lack of reliable, data-driven guidelines has contributed to opioid diversion and increased addiction risks among patients. Thus, there is an urgent need to forecast opioid prescription quantities for surgical patients and measure the associated uncertainty to establish better guidelines for opioid prescriptions. Second, lung cancer remains a major and deadly threat, claiming approximately 150,000 lives annually in the United States, underscoring the need for more effective intervention strategies. Recently, there has been growing interest in understanding the implications of lung muscle metrics on lung cancer mortality. This effort has the potential to lead to personalized interventions that encompass clinical, nutritional, and physical aspects tailored to individual patient profiles. The proposal is motivated by two large databases in which Principal Investigator Dr. Yi Li is actively involved. First, the Michigan Surgical Quality Collaborative (MSQC) project includes data from 21,033 opioid-naïve adult postoperative patients across 70 hospitals, capturing demographic, perioperative, clinical, and mortality characteristics through postoperative surveys. Patients who underwent surgery in Michigan between January 2017 and October 2019 were prescribed opioids at discharge, with ongoing use or refill determined by their pain levels. Second, the Boston Lung Cancer Survival Cohort (BLCSC) study, one of the largest lung cancer cohorts in the country, has enrolled 12,951 lung cancer cases at Massachusetts General Hospital and Dana-Farber Cancer Institute since 1992. It collects comprehensive demographic, smoking, and dietary information, along with pathology, imaging, treatment details, oncogenic mutation status, serum, white blood cells, germline DNA, and tumor tissues. Leveraging these rich databases, we aim to develop new methods to predict and infer adverse outcomes. We propose various methods to predict prescribed opioid dose levels post-surgery, quantify the associated uncertainty, and analyze refill patterns for individual patients. Additionally, we will develop methods to elucidate the impacts of lung muscle metrics on lung cancer mortality, which could have significant interventional implications. In summary, these rich databases and new methodologies will enable our findings to inform individual-based prescription guidelines, detect new prognostic biomarkers and impact medical practice.





None