Harnessing the Power of AI for Cancer Detection

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 system for early-stage detection of esophageal cancers during routine endoscopy, a machine learning algorithm to identify patients in public health records with a likelihood of having common variable immunodeficiency, artificial intelligence (AI) models to identify patients at risk of developing pancreatic cancer, and more!

Esophageal cancer: A deep learning system has enhanced early-stage detection of esophageal cancers during routine endoscopy, according to a randomized controlled trial recently published in Science Translational Medicine. The AI system ENDOANGEL-ELD nearly doubled clinicians’ ability to identify high-risk esophageal lesions (HrELs), including cancer and precancerous conditions, compared to unassisted endoscopy. This improvement led to detecting one additional HrEL case per 111 patients screened. Conducted by Dr. Shao-Wei Li and colleagues at Taizhou Hospital in China, this is the first large-scale randomized controlled trial (RCT) validating the efficacy of a deep learning–based system for esophageal cancer detection.

The trial involved 3,117 patients randomly assigned to either convolutional neural network (CNN)–assisted or unassisted endoscopy. Results showed a significant increase in HrEL detection rates with AI assistance, achieving 1.8% compared to 0.9% in the control group. The system demonstrated high sensitivity (89.7%), specificity (98.5%), and accuracy (98.2%), with no adverse events reported. These findings suggest that AI can improve early diagnosis and treatment of esophageal cancer, potentially enhancing patient outcomes.

Diagnosing rare disease: Human inborn errors of immunity include rare disorders with functional and quantitative antibody deficiencies, such as common variable immunodeficiency (CVID). Due to the rarity and heterogeneity of CVID, patients often face delayed diagnoses and treatments, ranging from five to 15 years after symptom onset. The lack of a single causal mechanism means no definitive genetic tests are available.

Johnson et al developed PheNet, a machine learning algorithm that identifies CVID patients from electronic health records (EHRs). The research was recently published in Science Translational Medicine. PheNet uses phecodes and IgG laboratory test results to rank patients by their likelihood of having CVID. Trained on 197 verified cases, PheNet identified 74% of the top 100 ranked patients as highly-probable CVID cases in a large EHR dataset from the University of California, Los Angeles (UCLA) Health. It could have diagnosed over half of the known CVID patients one or more years earlier. Further validation on over six million records from various medical systems demonstrated its effectiveness. This study highlights AI’s potential in expediting rare disease diagnosis.

Pancreatic cancer: Current surveillance guidelines for pancreatic cancer are limited to high-risk individuals (HRIs), who make up only 20% to 25% of cases. In a Perspective piece in Lancet, Dr. Anirban Maitra and Dr. Eric Topol explore how AI can enhance early detection beyond this limited scope. Recent studies highlight the potential of transformer AI models to identify a broader population at “sporadic” risk of developing pancreatic cancer within 12 months using longitudinal EHR data. These models consistently pinpoint diabetes as a significant predictor of future pancreatic cancer. The authors advocate for developing multimodal AI models that integrate genomic data, gut microbiome information, and unstructured EHRs to better stratify risk in the general population. This integration could improve early detection, thereby enhancing patient outcomes. Pancreatic cancer’s rising global incidence and typical late-stage diagnosis underscore the urgency of these advancements.

About Dr. Haque

Waqas Haque, MD, MPH, is an incoming Hematology/Oncology Fellow at the University of Chicago. He is currently finishing up his Internal Medicine Residency at New York University (NYU) in a Clinical Investigator Track. As a Clinical Investigator Track Resident, Dr. Haque has balanced his patient care work with a variety of research projects. During his fellowship training at University of Chicago, he plans to further his work in innovative clinical trial design, value-based care delivery to cancer patients, and clinical investigation.

Banner ad

Related Articles

Responses

Your email address will not be published. Required fields are marked *