Search

Deep neural networks (DNNs) have not been proven to detect blood loss (BL) or predict surgeon performance from video.To train a DNN using video from cadaveric training exercises of surgeons controlling simulated internal carotid hemorrhage to predict clinically relevant outcomes.Video was input as a series of images; deep learning networks were developed, which predicted BL and task success from images alone (automated model) and images plus human-labeled instrument annotations (semiautomated model). These models were compared against 2 reference models, which used average BL across all trials as its prediction (control 1) and a linear regression with time to hemostasis (a metric with known association with BL) as input (control 2). The root-mean-square error (RMSE) and correlation coefficients were used to compare the models; lower RMSE indicates superior performance.One hundred forty-three trials were used (123 for training and 20 for testing). Deep learning models outperformed controls (control 1: RMSE 489 mL, control 2: RMSE 431 mL, R2 = 0.35) at BL prediction. The automated model predicted BL with an RMSE of 358 mL (R2 = 0.4) and correctly classified outcome in 85% of trials. The RMSE and classification performance of the semiautomated model improved to 260 mL and 90%, respectively.BL and task outcome classification are important components of an automated assessment of surgical performance. DNNs can predict BL and outcome of hemorrhage control from video alone; their performance is improved with surgical instrument presence data. The generalizability of DNNs trained on hemorrhage control tasks should be investigated.

BACKGROUND Hospital readmissions are commonly linked to elevated health care costs, with significant financial incentive introduced by the Affordable Care Act to reduce readmissions. OBJECTIVE To study the association between patient, hospital, and payer factors with national rate of readmission in acoustic neuroma surgery. METHODS All adult inpatients undergoing surgery for acoustic neuroma in the newly introduced Nationwide Readmissions Database from 2013 to 2014 were included. We identified readmissions for any cause with a primary diagnosis of neurological, surgical, or systemic complication within 30- and 90-d after undergoing acoustic neuroma surgery. Multivariable models were employed to identify patient, hospital, and administrative factors associated with readmission. Hospital volume was measured as the number of cases per year. RESULTS We included patients representing a weighted estimate of 4890 admissions for acoustic neuroma surgery in 2013 and 2014, with 355 30-d (7.7%) and 341 90-d (9.1%) readmissions. After controlling for patient, hospital, and payer factors, procedural volume was significantly associated with 30-d readmission rate (OR [odds ratio] 0.992, p = 0.03), and 90-d readmission rate (OR 0.994, p = 0.047). The most common diagnoses during readmission in both 30- and 90-d cohorts included general central nervous system complications/deficits, hydrocephalus, infection, and leakage of cerebrospinal fluid (rhinorrhea/otorrhea). CONCLUSION After controlling for patient, hospital, and payer factors, increased procedural volume is associated with decreased 30- and 90-d readmission rate for acoustic neuroma surgery. Future studies seeking to improve outcomes and reduce cost in acoustic neuroma surgery may seek to further evaluate the role of hospital procedural volume and experience.

Background Atypical pituitary adenomas (APAs) are a subset of pituitary adenomas (PAs) characterized by the 2004 World Health Organization (WHO) guidelines to have higher risk histopathological features than typical PAs. In July 2017, the WHO published an update to their classification of pituitary tumors and abandoned the APA terminology. Objective To assess the prevalence and outcomes of patients diagnosed with APA through a literature review. Focus was placed on variation in the application of the previous WHO criteria and on rates of recurrence. Methods A systematic review of PubMed (2004-July 2017) was performed to identify studies reporting prevalence and clinical characteristics/outcomes of APA. Eight studies were analyzed for prevalence. Six studies reporting histopathological details were analyzed in depth. Results Of the 7105 included patients, 373 (5.2%) met criteria for APA (prevalence range: 3%-15%). Only 2 of 8 studies utilized identical grading criteria, demonstrating a lack of standardized application. Most APAs (84%) were macroadenomas, with 52% invasive on magnetic resonance imaging. Nonfunctional PAs were most common (37%), followed by prolactinomas (23%) and Growth Hormone adenomas (21%). Recurrence/progression occurred in 21% of APA patients (follow-up range 37-75 mo). Only 2 of 8 studies reported an association between APA diagnosis and recurrence/progression. Conclusion Based on diagnostic variability and lack of association with clinical outcomes, refinement of criteria for APA was necessary. The WHO update eliminates the ambiguity in APA diagnosis in favor of criteria that emphasize clinical behavior (invasion, recurrence, and resistance to treatment) and molecular markers. Our review supports abandonment of the previous APA designation due to limited prognostic utility.