AI Model Automates Cancer Severity Assessment from Pathology Reports, Accelerating Clinical Trial Enrollment
核心洞察
Researchers have developed an AI model, BB-TEN (搜索), that automates cancer (搜索) staging using pathology reports, significantly reducing the time required for clinical trial candidate selection.
The AI model utilizes a transformer-based approach, trained on a large dataset of pathology reports from The Cancer (搜索) Genome Atlas, demonstrating high accuracy across various cancer types.
This AI tool can be generalized to other institutions without extensive retraining, streamlining patient classification for clinical trials, observational studies, and potential treatments.
An artificial intelligence (AI) method has been developed to streamline cancer (搜索) clinical trials by automating the classification of patients based on cancer severity. Led by Cedars-Sinai investigators, the new AI model uses patients’ pathology reports to expedite the selection process for clinical trial candidates, potentially saving months of delay. The study, published in Nature Communications, highlights the model's ability to extract and interpret textual information from pathology reports, accurately staging cancers and offering a significant advancement over traditional tumor registries.
Automating Cancer Staging with AI
Traditional tumor registries, which rely on manual review of laboratory reports and clinical notes, can be slow and resource-intensive. "By the time a cancer (搜索) patient’s data is entered into a tumor registry, months may have passed, along with the opportunity for the patient to participate in relevant clinical trials or other treatments," said Nicholas Tatonetti, PhD, vice chair of Computational Biomedicine at Cedars-Sinai and corresponding author of the study. The AI model addresses this issue by rapidly identifying cancer stages directly from pathology reports, significantly reducing delays and expanding patient access to clinical trials.
How the AI Model Works
The AI model, named BB-TEN (搜索) (Big Bird – TNM staging Extracted from Notes), is based on a transformer model designed to mimic human decision-making. The model was initially trained using publicly available pathology reports from The Cancer (搜索) Genome Atlas, encompassing nearly 7,000 patients and 23 cancer types. To validate its performance, the model was then applied to nearly 8,000 pathology reports from a single medical center, demonstrating high accuracy as measured by standard AI evaluation metrics.
"This was an important finding because it means that our AI model is an 'off-the-shelf' tool that can be generalized to other institutions without requiring that it be trained for each location," Tatonetti explained. The model's ability to generalize across different institutions enhances its potential for widespread adoption and impact.
Applications Beyond Clinical Trials
In addition to screening patients for clinical trials, the AI model can automate patient classification for observational and retrospective data analysis, as well as potential treatments. According to Tatonetti, future research could integrate pathology text with other clinical data to further personalize cancer (搜索) treatment. The investigators have made the BB-TEN (搜索) model available to other institutions for academic uses, promoting further innovation and collaboration in the field.
Expert Commentary
"By speeding up the selection of candidates for cancer (搜索) clinical trials, this innovative AI model shows promise for accelerating the development of relevant treatments and making them available to more patients," said Jason Moore, PhD, chair of the Department of Computational Biomedicine at Cedars-Sinai.