Optimizing Individualized Colorectal Cancer Treatment and Prognostic Prediction via Causal Machine Learning
Abstract
Colorectal cancer affects 5.25 million people worldwide. Pathology evaluation (i.e., visual assessment of the cancer samples under the microscope) is the key to diagnosing this deadly disease. However, this standard microscopic evaluation cannot predict which patients will respond to treatments (such as immunotherapy, a type of treatment that reactivates the immune system against cancer cells) or develop undesirable side effects. The scientific objective of this proposal is to develop new artificial intelligence (AI) methods to analyze the microscopic images of colorectal cancer samples and predict treatment responses and side effects. The rationale is that AI methods can automatically identify hidden signals from a large amount of microscopic imaging data and discover previously unknown relationships from the data. We will combine AI methods with causal modeling approaches to enable reliable clinical prediction and optimize treatments for each patient. Dr. Yu is the Principal Investigator (PI) of this proposal. His career goal is to develop reliable AI methods for analyzing high-resolution microscopic images of colorectal cancer samples. This award will significantly advance Dr. Yu’s career in colorectal cancer research (a fiscal year 2022 FY22 Peer Reviewed Cancer Research Program PRCRP Topic Area) by providing him with the time and resources to execute the proposed research project. The proposed study will enable accurate treatment response prediction and facilitate treatment optimization for colorectal cancer patients, which will address two of the FY22 PRCRP Overarching Challenges (Diagnostics/Prognostics: Identify strategies to predict treatment resistance, recurrence, and the development of advanced disease and Therapeutics: Evaluation from longitudinal collection of deep multidimensional characterization of clinically annotated research biospecimens during disease progression and/or treatment). The proposed research and career development plan (joint lab meetings with well-established colorectal cancer researchers, networking with leaders of the Virtual Cancer Center, and additional training in grant writing) will provide Dr. Yu with additional exposure to cutting-edge research ideas and support the PI in becoming a leader in colorectal cancer pathology research. The ultimate applicability of the research is to establish reliable methods for predicting the treatment response and side effects of immunotherapy among colorectal cancer patients. Results from this study will assist colorectal cancer patients in choosing the right treatments according to their individual differences. Potential clinical applications include informing each patient and their clinicians about the most likely treatment responses and side effects before initiating any treatments, which will allow patients and clinicians to make an informed joint decision on the treatment strategies. These results will benefit patients by selecting the optimal treatments and avoiding ineffective medications for their cancers. The risk associated with uncertainty in the prediction will be mitigated through informed consent and educating clinicians and patients regarding the inherent uncertainty in medical practice. We expect to complete the development and evaluation of our prediction models in 4 years using three large patient populations. Our study will advance the field of cancer pathology research by establishing reliable methods for analyzing high-resolution digital pathology images, and it will advance patient care by providing colorectal cancer patients with accurate clinical predictions. The proposed research will benefit active-duty Service Members, Veterans, and other military beneficiaries by enabling fast and accurate clinical prediction using pathology samples. Because the patient population of colorectal cancer is getting younger in recent decades (12.1% of colorectal cancer patients in the U.S. are under 50 years old i
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 04, 2024
- Source ID
- HT94252310523
Entities
People
- Kun-Hsing Yu
Organizations
- Harvard University
- United States Army