Advanced machine-learning techniques successfully identified key predictors of intestinal resection in incarcerated inguinal hernia patients, including peritonitis, intestinal obstruction, neutrophil count, C-reactive protein, and preoperative total protein. The constructed model, validated externally, demonstrated strong predictive performance with an area under the curve exceeding 0.8 for all ten algorithms tested. Notably, the k-nearest neighbor algorithm showed […]
Category: Digital Surgery and Telemedicine
Machine learning predicts duodenal stump leakage in gastric cancer.
A machine learning model was developed to predict duodenal stump leakage (DSL) in 1,107 gastric cancer patients post-gastrectomy. The model utilized 189 clinical features and evaluated six algorithms, with extreme gradient boosting (XGB) achieving the highest area under the receiver operating characteristic curve (AUROC) score of 0.880. Random forest followed with a score of 0.858. […]
Gradient boosting model predicts liver metastasis in pancreatic neuroendocrine tumors.
A study utilizing data from 7,463 pancreatic neuroendocrine tumor (PanET) patients developed a gradient boosting machine (GBM) model, marking it as the most effective tool for predicting liver metastasis. The model achieved an AUC of 0.937 and an accuracy of 0.87, highlighting independent risk factors like T-stage, N-stage, and previous treatments. It culminated in a […]
Multimodal system enhances diagnostic accuracy for hepatic conditions
A deep learning diagnostic system combining CT and ultrasound achieved superior accuracy in detecting hepatic echinococcosis and other liver conditions compared to individual modalities and physician evaluations. The integrated model demonstrated an impressive accuracy of 81.6% in internal tests and 84.9% in external assessments, with high sensitivity and specificity. This innovation significantly reduces misdiagnosis rates, […]
AI aligns closely with colorectal cancer treatment decisions
A retrospective study evaluated the concordance between therapeutic recommendations made by multidisciplinary teams and those generated by ChatGPT for colorectal cancer. Of 100 patients, pre-therapeutic discussions showed a 72.5% complete concordance, while post-therapeutic discussions revealed an increase to 82.8%. Discordance was notably higher among patients over 77 years old and with higher ASA classifications. These […]
Multimodal ML Model Improves Delirium Detection Rates
A novel multimodal machine learning model significantly enhanced delirium risk stratification in hospitalized older adults. Validation outcomes revealed an impressive area under the curve of 0.94, with monthly delirium detection rates soaring from 4.42% to 17.17% following model deployment. Moreover, the post-deployment cohort experienced reduced daily doses of benzodiazepines and olanzapine, indicating potential improvements in […]
Machine learning model predicts anastomotic strictures post-surgery
A novel machine learning model demonstrates the potential to predict anastomotic strictures in patients following esophageal cancer surgery. Analyzing data from 1,549 patients, the gradient boosting machine achieved an area under the curve (AUC) of 0.886 in the training set and 0.872 in the validation set. Key predictive variables included anastomotic leakage, suture method, and […]
Deep learning model achieves high accuracy in ureter identification
A deep learning computer vision model demonstrated precise real-time identification of the left ureter during laparoscopic sigmoidectomy, achieving a mean average precision of 0.92. Key evaluation metrics, including precision, recall, and dice coefficient, reached impressive values of 0.94, 0.88, and 0.90, respectively. Operating at 32 frames per second, the model significantly aids surgical navigation. Despite […]
ChatGPT-4o is superior in aiding gastric cancer decisions
ChatGPT-4o significantly outperformed Gemini Advanced in generating treatment recommendations for advanced gastric cancer, as evidenced by a structured evaluation of responses to ten clinical questions. It provided superior recommendations in surgical suggestions and chemotherapy options during multidisciplinary team discussions. Additionally, it excelled in analyzing rare cases from PubMed, demonstrating increased accuracy and consistent evaluator agreement. […]
Deep learning effectively identifies pathologic complete response in esophageal cancer.
A deep neural network-based endoscopic evaluation accurately detected pathologic complete response (PCR) in esophageal cancer patients after neoadjuvant chemotherapy, achieving median sensitivity, specificity, and accuracy rates of 80%, 90%, and 85%, respectively. Among 1041 patients, 354 demonstrated PCR, correlating with significantly improved overall and recurrence-free survival. The findings suggest the potential for this AI technology […]