Updates on AI and Deep Learning for Cancer Detection and Outcomes Prediction

In this latest edition of The Way Ahead: The Convergence of Technology and Cancer Care, Dr. Waqas Haque shares his perspectives on a deep learning model for gallbladder cancer detection, a precision medicine platform for pancreatic ductal adenocarcinoma outcome prediction, and more! 

Artificial intelligence corner: Gallbladder cancer (GBC) poses a diagnostic challenge due to its aggressiveness and imaging similarities with benign lesions, and delayed diagnosis harms survival rates for patients with GBC. An article published in Lancet: Regional Health outlines a prospective study at a North Indian tertiary care hospital where a deep learning (DL) model was trained and validated for automatic GBC detection via abdominal ultrasound. After being trained on data from 2019 to 2021 with 233 patients in the training set, the DL model featured comparable diagnostic performance to experienced radiologists in an independent test cohort of 273 patients. The DL model had greater sensitivity than a radiologist (88% vs 73%) for mural thickening GBC and better overall performance for different GBC situations, including smaller lesions and the presence of stones. While the DL-based approach shows promise, the authors suggest multicenter studies to fully explore its role in GBC diagnosis and under-resourced settings.  

Predicting survival: In pancreatic ductal adenocarcinoma (PDAC), serum biomarker CA 19-9 is the only FDA-approved biomarker commonly used for PDAC prognostication, despite limitations including a high false-positive rate. The authors of a Nature Cancer study published last month present the Molecular Twin, a precision medicine platform claiming to make a bioinformatic, virtual replica of a patient for longitudinal analysis. Analyzing data from 6,363 PDAC patients, the Molecular Twin was able to accurately predict disease survival, outperforming existing methods. After combing through tumor specimen data, demographics and clinical characteristics, and computational pathology information, plasma proteins were identified as the top single-omic predictor of disease survival. A parsimonious model with 589 multi-omic features demonstrated comparable predictive power. This innovative platform not only refines biomarker panels but also enhances outcome prediction models for PDAC.  

Clinical trials corner: Earlier this year, we spoke with OncoHost CEO Dr. Ofer Sharon, who is leading the precision medicine company’s efforts to validate a machine learning algorithm to predict immunotherapy-related adverse effects through biomarker data. The company’s PROphet platform will be utilized to assess blood samples in an open-label phase 3 trial of serplulimab (NCT05468489) for response and adverse events prediction. You can find more information here.

About Dr. Haque

Waqas Haque, MD, MPH, is a third-year Internal Medicine Resident at New York University (NYU) in a Clinical Investigator Track. He recently matched to the University of Chicago for fellowship, which he will be beginning later in 2024. As a Clinical Investigator Track Resident, Dr. Haque has balanced his patient care work with a variety of research projects. He hopes to begin fellowship training next year in Medical Hematology/Oncology at an academic program with opportunities to further his work in innovative clinical trial design, value-based care delivery to cancer patients, and becoming an early-stage clinical investigator.

For More Information

 

Gupta P, Basu S, Rana P, et al (2024). Deep-learning enabled ultrasound based detection of gallbladder cancer in northern India: a prospective diagnostic study. Lancet Reg Health Southeast Asia, 100279. [Epub ahead of print] DOI:10.1016/j.lansea.2023.100279

 

Osipov A, Nikolic O, Gertych A, et al (2024). The Molecular Twin artificial-intelligence platform integrates multi-omic data to predict outcomes for pancreatic adenocarcinoma patients. Nature Cancer. [Epub ahead of print] DOI:10.1038/s43018-023-00697-7

Clinical Trials Arena (2024). Henlius and OncoHost partner to predict trial outcomes using machine learning. Available at: https://www.clinicaltrialsarena.com/news/henlius-and-oncohost-partner-to-predict-trial-outcomes-using-machine-learning/?cf-view

Related Articles