Machine learning and survival analysis in gallbladder adenocarcinoma (GBA): Key findings.

e16483 Background: GBA accounts for 90% of gallbladder cancers and is challenging to manage owing to its late presentation and varied treatment approaches. This study is the first to apply machine learning (ML) to improve survival by identifying key prognostic factors. Methods: Data were obtained from the SEER database (2004-2021). Patients who met any of the following criteria were excluded: diagnosis not confirmed by histology, previous history of cancer or other concurrent malignancies, or unknown data. To identify prognostic variables, we conducted Cox regression analysis and constructed prognostic models using ML algorithms to predict the 5-year survival. Patient records were randomly divided into training (70%) and validation (30%) sets. A validation method incorporating the area under the curve (AUC) of the receiver operating characteristic curve was used to validate the accuracy and reliability of the ML models. We also investigated the role of multiple therapeutic options using Kaplan-Meier survival analysis. Results: A total of 12,890 patients were included. Most patients (71%) were female and 51.1% were white, followed by Asian or Pacific Islander patients (14.8%). The median patient age was 70 years, and the median tumor size was 3 cm. Most patients were classified as T2 (37.6%) or T3 (40.9%), with T4 being rare (4.2%). Nodal involvement was observed in 68.9% of cases, while M0 accounted for 76.4%. Of the metastatic cases, 9.2% were in the liver and 1.1% were in the liver and lungs. Among the patients, 86.3% underwent surgery, and only 39.6% received chemotherapy. Patients aged < 70 years had higher 5-year OS (31%) and CSS (34.1%) than those aged ≥ 70 years (OS: 18.8%, CSS: 26.3%). Asians had the highest survival rate (OS, 31.4%; CSS, 36.7%). Patients who received chemotherapy had lower survival rates than those who did not receive chemotherapy (OS: 19.1% vs. 28.1%). Multivariate Cox regression analysis identified older age, male sex, metastasis, and White and Black race as poor prognostic factors, whereas surgery was a good prognostic factor. Gradient boosting and Random Forest are the most accurate models. The ML models identified TNM stage as the most significant prognostic factor, followed by age and race. The performance metrics for all ML algorithms are summarized in Table. Conclusions: This study presents a novel and practical tool for the prognosis and management of GBA that offers valuable support for personalized clinical decisions. However, further research is required to clarify the role of adjuvant therapy in treatment and its impact on patient outcomes. ML Model Accuracy Precision Recall F1 score AUC LR 84.7% 86.73% 96.7% 91.4% 0.821 KNN 82.1% 87.1% 92.4% 89.7% 0.757 RFC 85.7% 87.6% 96.6% 91.9% 0.865 GBC 85.% 86.71% 97.1% 91.6% 0.826 MLP 84.88% 88.44% 94.37% 91.31% 0.848