RESUMO
This study used artificial intelligence techniques to identify clinical cancer biomarkers for recurrent gastric cancer survivors. From a hospital-based cancer registry database in Taiwan, the datasets of the incidence of recurrence and clinical risk features were included in 2476 gastric cancer survivors. We benchmarked Random Forest using MLP, C4.5, AdaBoost, and Bagging algorithms on metrics and leveraged the synthetic minority oversampling technique (SMOTE) for imbalanced dataset issues, cost-sensitive learning for risk assessment, and SHapley Additive exPlanations (SHAPs) for feature importance analysis in this study. Our proposed Random Forest outperformed the other models with an accuracy of 87.9%, a recall rate of 90.5%, an accuracy rate of 86%, and an F1 of 88.2% on the recurrent category by a 10-fold cross-validation in a balanced dataset. We identified clinical features of recurrent gastric cancer, which are the top five features, stage, number of regional lymph node involvement, Helicobacter pylori, BMI (body mass index), and gender; these features significantly affect the prediction model's output and are worth paying attention to in the following causal effect analysis. Using an artificial intelligence model, the risk factors for recurrent gastric cancer could be identified and cost-effectively ranked according to their feature importance. In addition, they should be crucial clinical features to provide physicians with the knowledge to screen high-risk patients in gastric cancer survivors as well.
RESUMO
PURPOSE: The purpose of this study was to compare health-related quality of life (HRQoL) and costs associated with 2 adjuvant chemotherapy regimens [capecitabine-based therapy versus 5-fluorouracil/leucovorin (5-FU/LV)-based therapy] in stage III colorectal cancer patients. METHODS: We conducted a prospective, open-label, observational, multicenter study from July 2008 to July 2011. The European Organization for Research and Treatment of Cancer QLQ-C30 and QLQ-CR38 questionnaires was used to assess HRQoL before, during, and after treatment. The direct and indirect costs of adjuvant treatment were estimated from a specially prepared questionnaire, the National Health Insurance Research Database, and other published sources. We used propensity scoring to match samples between groups and performed multivariate analyses to adjust for differences in patient demographics and clinical characteristics. RESULTS: A total of 497 patients were enrolled, and 356 completed the surveys. Following propensity score matching, 239 patients were included in the analysis (122 in the capecitabine-based group, 117 in the 5-FU/LV-based group). Global HRQoL scores did not differ significantly between the two groups. However, compared to patients in the 5-FU/LV-based group, patients in the capecitabine-based group had less nausea and vomiting (mid-term, P = 0.024; final, P = 0.013), appetite loss (mid-term, P < 0.0001; final, P = 0.001), and fewer side effects from chemotherapy (mid-term, P = 0.017). In addition, the monthly cost of capecitabine-based therapy was lower than those of 5-FU/LV-based therapy [NT$31,895.46 (US$1063.18) vs. NT$79,159.24 (US$2638.64) per patient]. CONCLUSIONS: Capecitabine is a reasonable alternative and cost-effective treatment option under current conditions for patients with stage III colorectal cancer.