Comparing Frailty Indices for Risk Stratification in Urologic Oncology: Which Index to Choose?

OBJECTIVE: To compare the predictive ability of the modified Frailty Index (mFI) and the revised Risk Analysis Index (RAI-Rev) for perioperative outcomes in patients undergoing major urologic oncologic surgery, aiming to identify the optimal frailty screening tool for surgical risk stratification. METHODS: NSQIP was queried to identify patients undergoing radical prostatectomy, partial or radical nephrectomy, or radical cystectomy between 2013 and 2017. We investigated the association of mFI and RAI-Rev with the following 30-day perioperative outcomes using multivariable logistic regression: major complications, Clavien grade ≥4 complications, non-home discharge, 30-day readmission, and all-cause mortality. Receiver-operating characteristic curve analysis compared the predictive performances of the 2 frailty instruments, with differences between the C-statistics assessed using DeLong's test. RESULTS: Among 101,739 patients, 30-day major complication rates varied from 2.40% in prostatectomy to 26.86% in cystectomy, non-home discharge rates ranged from 1.92% to 13.54%, and mortality rates were between 0.16% and 1.43%. RAI-Rev showed higher discriminatory ability for mortality (C-statistic: 0.688-0.798) and non-home discharge (C-statistic: 0.638-0.734) compared to mFI (C-statistic: 0.594-0.677 and 0.593-0.639, respectively). Both frailty indices had similar discriminatory ability for major perioperative complications (C-statistic: 0.531-0.607). DeLong's test confirmed statistically significant differences in C-statistics between RAI-Rev and mFI for mortality (P <.001) and non-home discharge (P <.001) across all surgical cohorts. CONCLUSION: RAI-Rev may have greater utility as a frailty prognostic tool than mFI among patients undergoing major urologic surgery. Prospective studies and clinical trials exploring frailty should consider these results during trial design.

Artificial intelligence model for automated surgical instrument detection and counting: an experimental proof-of-concept study.

BACKGROUND: Retained surgical items (RSI) are preventable events that pose a significant risk to patient safety. Current strategies for preventing RSIs rely heavily on manual instrument counting methods, which are prone to human error. This study evaluates the feasibility and performance of a deep learning-based computer vision model for automated surgical tool detection and counting. METHODS: A novel dataset of 1,004 images containing 13,213 surgical tools across 11 categories was developed. The dataset was split into training, validation, and test sets at a 60:20:20 ratio. An artificial intelligence (AI) model was trained on the dataset, and the model's performance was evaluated using standard object detection metrics, including precision and recall. To simulate a real-world surgical setting, model performance was also evaluated in a dynamic surgical video of instruments being moved in real-time. RESULTS: The model demonstrated high precision (98.5%) and recall (99.9%) in distinguishing surgical tools from the background. It also exhibited excellent performance in differentiating between various surgical tools, with precision ranging from 94.0 to 100% and recall ranging from 97.1 to 100% across 11 tool categories. The model maintained strong performance on a subset of test images containing overlapping tools (precision range: 89.6-100%, and recall range 97.2-98.2%). In a real-time surgical video analysis, the model maintained a correct surgical tool count in all non-transition frames, with a median inference speed of 40.4 frames per second (interquartile range: 4.9). CONCLUSION: This study demonstrates that using a deep learning-based computer vision model for automated surgical tool detection and counting is feasible. The model's high precision and real-time inference capabilities highlight its potential to serve as an AI safeguard to potentially improve patient safety and reduce manual burden on surgical staff. Further validation in clinical settings is warranted.

Automated surgical step recognition in transurethral bladder tumor resection using artificial intelligence: transfer learning across surgical modalities.

Objective: Automated surgical step recognition (SSR) using AI has been a catalyst in the "digitization" of surgery. However, progress has been limited to laparoscopy, with relatively few SSR tools in endoscopic surgery. This study aimed to create a SSR model for transurethral resection of bladder tumors (TURBT), leveraging a novel application of transfer learning to reduce video dataset requirements. Materials and methods: Retrospective surgical videos of TURBT were manually annotated with the following steps of surgery: primary endoscopic evaluation, resection of bladder tumor, and surface coagulation. Manually annotated videos were then utilized to train a novel AI computer vision algorithm to perform automated video annotation of TURBT surgical video, utilizing a transfer-learning technique to pre-train on laparoscopic procedures. Accuracy of AI SSR was determined by comparison to human annotations as the reference standard. Results: A total of 300 full-length TURBT videos (median 23.96 min; IQR 14.13-41.31 min) were manually annotated with sequential steps of surgery. One hundred and seventy-nine videos served as a training dataset for algorithm development, 44 for internal validation, and 77 as a separate test cohort for evaluating algorithm accuracy. Overall accuracy of AI video analysis was 89.6%. Model accuracy was highest for the primary endoscopic evaluation step (98.2%) and lowest for the surface coagulation step (82.7%). Conclusion: We developed a fully automated computer vision algorithm for high-accuracy annotation of TURBT surgical videos. This represents the first application of transfer-learning from laparoscopy-based computer vision models into surgical endoscopy, demonstrating the promise of this approach in adapting to new procedure types.

Frailty and survival among veterans treated with abiraterone or enzalutamide for metastatic castration-resistant prostate cancer.

INTRODUCTION: Abiraterone and enzalutamide are treatments for metastatic castration-resistant prostate cancer (mCRPC). Due to a lack of head-to-head trials, they are prescribed interchangeably. However, the drugs have different pharmacokinetics and thus may have differing efficacy and adverse effects influenced by patient functional status and comorbid diseases. Additionally, mCRPC mainly affects older adults and since the prevalence of frailty increases with age, frailty is an important patient factor to consider in personalizing drug selection. MATERIALS AND METHODS: We conducted a retrospective observational study of US veterans treated with abiraterone or enzalutamide for mCRPC from September 2014 to June 2017. Frailty was assessed using the Veterans Affairs Frailty Index (VA-FI), which utilizes administrative codes to assign a standardized frailty score. Patients were categorized as frail if VA-FI scores were > 0.2. The primary outcome was difference in overall survival (OS) between the two treatment groups. Cox regression modeling and propensity score matching was used to compare between abiraterone and enzalutamide treatments. RESULTS: We identified 5,822 veterans, 57% of whom were initially treated with abiraterone and 43% with enzalutamide. Frail patients (n = 2,314; 39.7%) were older, with a mean age of 76.1 versus 74.9 years in the non-frail group (n = 3,508; 60.3%, p < 0.001) and had shorter OS compared to non-frail patients regardless of treatment group (18.5 vs. 26.6 months, p < 0.001). Among non-frail patients there was no significant difference in OS between abiraterone and enzalutamide treatment (27.7 vs 26.1 months, p = 0.07). However, frail patients treated with enzalutamide versus abiraterone had improved OS (20.7 vs 17.2 months, p < 0.001). In a propensity score matched analysis of frail patients (n = 2,070), enzalutamide was associated with greater median OS (24.1 vs 20.9 months, p < 0.001). In patients with dementia, enzalutamide was associated with longer OS (19.4 vs. 16.6 months, p = 0.003). DISCUSSION: In this study of 5822 US veterans with mCRPC, treatment with enzalutamide was associated with improved OS compared to abiraterone among frail veterans and veterans with dementia, but not among non-frail veterans. Future studies should evaluate interactions between frailty and cancer treatments to optimize selection of therapy among frail adults.