Frontiers in surgery · 2026
Comparative analysis of robotic vs. laparoscopic right hepatectomy: propensity score matching and machine learning analysis for outcome prediction.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Mazzotta Alessandro D, Samer Diab, Salina Giacomo, Ratti Francesca, Baldisseri Federico, Magistri Paolo, Belli Andrea, Ceccarelli Graziano, Izzo Francesco, Spampinato Marcello Giuseppe, De' Angelis Nicola, Pessaux Patrick, Piardi Tullio, Di Benedetto Fabrizio, Ammendola Michele, Aldrighetti Luca, Mennini Gianluca, Tedeschi Michele, Memeo Riccardo, Soubrane Olivier
Аннотация
BACKGROUND: Robotic right hepatectomy (RRH) is increasingly utilized as a minimally invasive liver surgery approach, yet its comparative effectiveness against laparoscopic right hepatectomy (LRH) remains debated. Complication prediction is crucial to efficiently choice the approach. This study aimed to compare RRH and LRH using a propensity score-matched (PSM) analysis and assess the predictive power of machine learning (ML), focusing on Random Forest-based feature selection.
METHODS: We retrospectively analyzed patients undergoing RRH and LRH, performing a 1:1 PSM based on preoperative characteristics. Postoperative outcomes were compared, and significant predictors of textbook outcomes (TO) were identified through univariate and multivariate analysis. A Random Forest model was developed, selecting key variables for TO achieving prediction.
RESULTS: After PSM, 30 RRH and 30 LRH patients were compared. RRH was associated with increased clamping (80% vs. 23.3%, p = 0.0001) and longer clamping duration (44 min vs. 35 min, p = 0.209) but resulted in a shorter hospital stay (6 vs. 7.5 days, p = 0.004) and fewer complications (13.3% vs. 53.3%, p = 0.001). Multivariate analysis identified Robotic surgical approach (OR 0.282, p = 0.004), Chronic hepatitis/Cirrhosis (OR 1.90, p = 0.034), and estimated blood loss (OR 2.001, p = 0.014) as significant predictors of TO. Feature selection via ML confirmed robotic surgical approach as one of the most relevant factors, alongside operative duration, estimated blood loss, and lesion size, in determining postoperative outcomes.
CONCLUSION: The choice of surgical approach is a key determinant of outcomes, as demonstrated by its significance in both multivariate analysis and ML-based feature selection. RRH shows potential advantages in reducing hospital stay and complications while requiring more intraoperative management. Machine learning, particularly Random Forest, enhances outcome prediction by identifying critical surgical factors, supporting a more tailored approach in hepatobiliary surgery.
METHODS: We retrospectively analyzed patients undergoing RRH and LRH, performing a 1:1 PSM based on preoperative characteristics. Postoperative outcomes were compared, and significant predictors of textbook outcomes (TO) were identified through univariate and multivariate analysis. A Random Forest model was developed, selecting key variables for TO achieving prediction.
RESULTS: After PSM, 30 RRH and 30 LRH patients were compared. RRH was associated with increased clamping (80% vs. 23.3%, p = 0.0001) and longer clamping duration (44 min vs. 35 min, p = 0.209) but resulted in a shorter hospital stay (6 vs. 7.5 days, p = 0.004) and fewer complications (13.3% vs. 53.3%, p = 0.001). Multivariate analysis identified Robotic surgical approach (OR 0.282, p = 0.004), Chronic hepatitis/Cirrhosis (OR 1.90, p = 0.034), and estimated blood loss (OR 2.001, p = 0.014) as significant predictors of TO. Feature selection via ML confirmed robotic surgical approach as one of the most relevant factors, alongside operative duration, estimated blood loss, and lesion size, in determining postoperative outcomes.
CONCLUSION: The choice of surgical approach is a key determinant of outcomes, as demonstrated by its significance in both multivariate analysis and ML-based feature selection. RRH shows potential advantages in reducing hospital stay and complications while requiring more intraoperative management. Machine learning, particularly Random Forest, enhances outcome prediction by identifying critical surgical factors, supporting a more tailored approach in hepatobiliary surgery.