Researchers trained an extreme gradient-boosting model, called the XGBoost model, to predict gastrointestinal stromal tumors (GISTs).
According to results published in BMC Gastroenterology, the features that best supported the diagnosis of GIST were round or round-like tumors with a computed tomography (CT) value of about 30 (25–37), delayed enhancement, and a liquid rather than calcific area inside the tumor.
The authors stated, “All the data we used to develop the model came from the patients’ preoperative clinical examinations and hematological tests, which would not cause any additional pain or economic stress for the patients.”
The model showed an accuracy of 0.73 (0.58–0.88) and a precision of 0.79 (0.60–0.95). Recall and f1-score were 0.87 (0.67–1.00) and 0.82 (0.70–0.92), respectively. The area under the receiver operating characteristic curve (AUROC) was 0.77 (0.57–0.90), and the C-index was 0.76 (0.56–0.89).
The most important predictors for GIST diagnosis in this study were the existence of a liquid area inside the tumor on endoscopic ultrasound (EUS) analysis, the ratio of the long and short diameters on CT analysis, the CT value of the tumor, enhancement of the tumor in the arterial and venous periods, and the existence of a calcific area inside the tumor on EUS analysis.
To develop the model, the authors used data retrospectively collected from 124 patients diagnosed with gastric GIST. The patients underwent partial gastrectomy and 90 individuals maintained the diagnosis of GIST postoperatively, while the remaining 34 were diagnosed with other gastrointestinal submucosal tumors.
These results require further validation in future large-scale, multicenter studies. In addition, the authors advised that their study was entirely based on data science and that the model was developed for patients whose initial diagnosis was gastric GIST.
Hu B, Wang C, Jiang K, et al. Development and validation of a novel diagnostic model for initially clinical diagnosed gastrointestinal stromal tumors using an extreme gradient-boosting machine. BMC Gastroenterol. 2021;21(1):481. doi:10.1186/s12876-021-02048-1