Cedars-Sinai researchers trained an AI to interpret pathology reports and translate them into machine-readable formats, making TCGA-Reports a pioneering dataset. This novel method solves the challenge of obtaining meaningful information from cancer patients’ pathology notes, which are generally PDF files of scanned papers.
Lead author Dr. Nicholas Tatonetti, PhD, noted the difficulties of making sense of pathologists’ PDF notes and the inability to enter them into computers. AI and OCR technology have enabled The Cancer Genome Atlas to publish machine-readable TCGA pathology reports.
Machine-readable pathology reports from over 10,000 cancer patients are in TCGA Reports. Tatonetti showed that the tool might help algorithms identify relevant pathology data to enable clinical trial recruitment and disease marker research more easily.
The researchers also noticed that their method might discover diseases in non-TCGA datasets. They then train the programs to distinguish cancer stages, enhancing accuracy. Cedars-Sinai Cancer Director Dan Theodorescu recommends mining physicians’ notes to gain the full patient picture, which aligns with precision medicine.
AI-powered progress will be a turning point in cancer research and clinical trials since it will provide data for researchers from different fields. Cedars-Sinai’s groundbreaking Molecular Twin Precision Oncology Platform is projected to benefit from this AI-based revolutionary method, which supports the institution’s objective to advance cancer research through cutting-edge technologies.
