TiME OUT: Time-specific machine-learning evaluation to optimize ultramassive transfusion.

BACKGROUND: Ultramassive transfusion (UMT) is a resource-demanding intervention for trauma patients in hemorrhagic shock, and associated mortality rates remains high. Current research has been unable to identify a transfusion ceiling or point where UMT transitions from lifesaving to futility. Furthermore, little consideration has been given to how time-specific patient data points impact decisions with ongoing high-volume resuscitation. Therefore, this study sought to use time-specific machine learning modeling to predict mortality and identify parameters associated with survivability in trauma patients undergoing UMT. METHODS: A retrospective review was conducted at a Level I trauma (2018-2021) and included trauma patients meeting criteria for UMT, defined as ≥20 red blood cell products within 24 hours of admission. Cross-sectional data were obtained from the blood bank and trauma registries, and time-specific data were obtained from the electronic medical record. Time-specific decision-tree models predicating mortality were generated and evaluated using area under the curve. RESULTS: In the 180 patients included, mortality rate was 40.5% at 48 hours and 52.2% overall. The deceased received significantly more blood products with a median of 71.5 total units compared with 55.5 in the survivors ( p < 0.001) and significantly greater rates of packed red blood cells and fresh frozen plasma at each time interval. Time-specific decision-tree models predicted mortality with an accuracy as high as 81%. In the early time intervals, hemodynamic stability, undergoing an emergency department thoracotomy, and injury severity were most predictive of survival, while, in the later intervals, markers of adequate resuscitation such as arterial pH and lactate level became more prominent. CONCLUSION: This study supports that the decision of "when to stop" in UMT resuscitation is not based exclusively on the number of units transfused but rather the complex integration of patient and time-specific data. Machine learning is an effective tool to investigate this concept, and further research is needed to refine and validate these time-specific decision-tree models. LEVEL OF EVIDENCE: Prognostic and Epidemiological; Level IV.

Impaired robotic surgical visualization: archaic issues in a modern operating room.

While robotic-assisted surgery (RAS) has been revolutionizing surgical procedures, it has various areas needing improvement, specifically in the visualization sector. Suboptimal vision due to lens occlusions has been a topic of concern in laparoscopic surgery but has not received much attention in robotic surgery. This study is one of the first to explore and quantify the degree of disruption encountered due to poor robotic visualization at a major academic center. In case observations across 28 RAS procedures in various specialties, any lens occlusions or "debris" events that appeared on the monitor displays and clinicians' reactions, the cause, and the location across the monitor for these events were recorded. Data were then assessed for any trends using analysis as described below. From around 44.33 h of RAS observation time, 163 debris events were recorded. 52.53% of case observation time was spent under a compromised visual field. In a subset of 15 cases, about 2.24% of the average observation time was spent cleaning the lens. Additionally, cautery was found to be the primary cause of lens occlusions and little variation was found within the spread of the debris across the monitor display. This study illustrates that in 6 (21.43%) of the cases, 90% of the observation time was spent under compromised visualization while only 2 (7.14%) of the cases had no debris or cleaning events. Additionally, we observed that cleaning the lens can be troublesome during the procedure, interrupting the operating room flow.

A Reliable and Sensitive Framework for Simultaneous Type and Stage Detection of Colorectal Cancer Polyps.

With the goal of enhancing the early diagnosis of colorectal cancer (CRC) polyps and reducing the risk of mortality in cancer patients, in this article, we present a unique diagnosis framework including a Vision-based Surface Tactile Sensor (VS-TS) and complementary Artificial Intelligence algorithms. Leveraging the morphological characteristics (i.e., shape and texture) and stiffness features of the CRC polyps, the proposed framework is able to reliably and sensitively identify their type and stage. To thoroughly characterize and identify the required VS-TS sensitivity for reliable identification of polyps, we first fabricated three different VS-TSs and qualitatively evaluated their performances on 48 different types of polyp phantoms fabricated based on four different types of realistic CRC polyps and three different materials. Next, to quantitatively compare the performance and sensitivity of the fabricated VS-TSs, we used Support Vector Machine (SVM) algorithm and employed various statistical metrics (i.e., accuracy, reliability, and sensitivity). Next, using the most sensitive VS-TS, we classified the type of tumors using the SVM algorithm and applied the t-Distributed Stochastic Neighbor Embedding algorithm to successfully identify the stiffness of classified polyp phantoms solely based on the output images of the VS-TS sensor. Results demonstrated that an SVM algorithm applied on the image outputs of a VS-TS with a Shore hardness of 00-40 scale is able to classify all types of polyps with > 90% accuracy, sensitivity, and reliability. We also repeated experiments on samples of ex-vivo lamb tripe tissues and successfully verified the high sensitivity and reliability of the proposed framework (i.e., > 94%).