Integration of machine learning into prognostic modeling for distal cholangiocarcinoma yielded a robust predictive model for disease-free survival after pancreaticoduodenectomy. Significant predictors of recurrence included lnr15, neural invasion, n stage, surgical radicality, and differentiation grade. The model demonstrated strong discriminatory capacity with internal and external validations showing high accuracy levels. This tool can enhance tailored […]
Category: Digital Surgery and Telemedicine
Artificial intelligence predicts non-transplantable recurrence in hepatocellular carcinoma
Artificial intelligence models accurately predicted non-transplantable recurrence in hepatocellular carcinoma patients post-hepatic resection. An ensemble AI model showed high accuracy (AUC 0.751-0.858) with incorporation of postoperative factors. Radiologic tumor burden and microvascular invasion were key predictors. Patients with predicted recurrence had lower survival rates, highlighting the models’ prognostic value. The AI models could guide treatment […]
Automatic Tissue Recognition in Surgery with Hyperspectral Imaging
Hyperspectral imaging, combined with machine learning, enables automatic tissue recognition during surgery. A study utilizing convolutional neural networks achieved high true positive rates for skin and liver tissues. Misclassifications mainly occurred between tissues with similar origins. The study underscores the potential of hyperspectral imaging in surgical optomics, offering a new approach for extracting tissue features. […]
Machine learning and deep learning in hernia surgery
Machine learning and deep learning technologies are increasingly utilized in hernia surgery, showing promise in predicting outcomes and identifying factors associated with postoperative complications. Thirteen studies from 2020 to 2023 were included, focusing on inguinal, ventral, and incisional hernias. These studies varied in population, types of models used, and outcomes predicted, showcasing the potential of […]
Machine Learning Model for Prediction of Permanent Stoma after Anterior Resection of Rectal Cancer
Retrospective study develops a machine learning model to predict permanent stoma formation post rectal cancer surgery. Utilizing support vector machine algorithm, the model shows high accuracy (AUC: 0.854) in training, testing, and external validation sets, aiding in preoperative patient stratification and personalized treatment plans. Results offer a practical tool for clinicians to enhance patient outcomes […]
Real-time Detection of Active Bleeding in Laparoscopic Colectomy using AI
Developing a real-time AI model to detect active intraoperative bleeding, this study achieved an average precision of 0.574 for active bleeding and 48.5 frames per second. Surgeons rated the sensitivity and overdetection areas of the model highly, indicating its potential to improve surgical outcomes by providing real-time support in laparoscopic colectomy. The AI model can […]
Effectiveness of Virtual Care in a Multidisciplinary Bariatric Program
Transition to virtual care in a bariatric program saw significant increase in virtual visits, with high patient satisfaction rates. Patient preferences varied by healthcare providers, with most patients preferring in-person visits with medical doctors. Despite the preference, virtual care offers added benefits and high satisfaction rates. The study highlights the importance of a hybrid model […]
Advancing Surgical Performance Assessment Through Digital Metrics
Surgeons evaluated 81 studies on digital metrics for nontechnical skills, highlighting diverse objective measures for cognitive load, situation awareness, and communication. Surgical sabermetrics, using sensors, offers promising outcomes for performance optimization. Gaps remain in assessing decision-making objectively. Future integration of physiological sensors could enhance holistic surgical performance evaluation and patient safety. Research Support, Non-U.S. Gov’t […]
Deep Learning Improves Prediction of Response to Neoadjuvant Chemotherapy in Colorectal Liver Metastasis
Deep learning models applied to prechemotherapy imaging accurately predict response to neoadjuvant chemotherapy in colorectal cancer liver metastases, outperforming clinical models. A study on 95 patients found the deep learning model had an AUC of 0.77, superior to the clinical model’s AUC of 0.41, showing potential to identify nonresponders and guide patient management towards curative […]
OR Black Box: A Game-Changer for Patient Safety and Training
Operating room black box technology allows for real-time monitoring of surgical procedures, improving transparency, and identifying areas for skill improvement. It enhances patient safety by providing a comprehensive analysis of clinical and non-clinical performances, offering a holistic view of intraoperative events. This innovative tool serves as a valuable teaching aid in surgical training, addressing limitations […]