Survival Prediction in Gallbladder Cancer Using CT Based Machine Learning
Problem
Doctors lacked a tool to predict overall survival for gallbladder cancer patients from pre-treatment CT scans. The study built a machine learning model combining CT features with clinical information to predict survival.
Why it matters
Gallbladder cancer is often lethal, and accurate prognosis can guide treatment decisions.
What the AI did
Radiomics features were extracted from manually segmented CT tumors using LIFEx; LASSO and Random Forest selected 15 features and built a survival risk model combined with clinical data.
Results
Hazard ratio 1.462 (95% CI: 1.014–2.107) for the CT-based model; 15 CT features selected; patients stratified into high/low-risk groups.
Impact
Offers a CT-based risk stratification for gallbladder cancer patients in a 141-patient retrospective cohort.
Record
| Venue | Frontiers in Oncology |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Zefan Liu, Guannan Zhu, Xian Jiang, Yunuo Zhao, Hao Zeng, Jing Jing, Xuelei Ma |
| Citations | 23 (OpenAlex, at import time) |
| References | 24 |
| Model or method | LASSO and Random Forest |
| Organization | not stated in source |
| Primary source | https://doi.org/10.3389/fonc.2020.604288 |
| Open access | gold |
| Keywords | gallbladder cancer, radiomics, CT, machine learning, survival prediction |
Importance rating
real but narrower advance
Prognosis tool for one cancer type; abstract only shows a 141-patient retrospective cohort, no broader scale stated.