A novel machine-learning algorithm for predicting mortality risk after hip fracture surgery. Issue 6 (June 2021)

• Record Type: Journal Article
• Authors: Li, Yi; Chen, Ming; Lv, Houchen; Yin, Pengbin; Zhang, Licheng; Tang, Peifu
• Is Part Of: Injury. Volume 52:Issue 6(2021)
• Page Start: 1487
• Page End: 1493
• Publication Date: 2021-06
• Subjects: Random forest -- Random survival forest -- Risk stratification model -- Hip fracture -- Mortality; Wounds and injuries -- Surgery -- Periodicals; Accidents -- Periodicals; Wounds and Injuries -- surgery -- Periodicals; Lésions et blessures -- Chirurgie -- Périodiques; Electronic journals; Electronic journals; 617.1
• DOI: 10.1016/j.injury.2020.12.008
• Languages: English
• ISSNs: 0020-1383

Abstract

Highlights: Novel risk stratification models for hip fracture patients to identify those with a higher mortality risk are built through a machine learning approach which has never been used in this population with this objective. Comparing with previously reported models, we achieve higher c-statistic values in both long-term and short-term mortality discrimination, which may imply a better performance of our model. The machine-learning algorithm enables us to acquire new knowledge on two aspects regarding the risk factors whichever identified exclusively in our study such as "RDW" or those previously reported such as "post-complication". These two aspects include firstly a horizontal comparison at a certain time point of the relative predictive values across risk factors; and secondly, a changing pattern description of each risk factor's contribution to mortality on temporal scales. Abstract: Introduction: Although several risk stratification models have been developed to predict hip fracture mortality, efforts are still being placed in this area. Our aim is to (1) construct a risk prediction model for long-term mortality after hip fracture utilizing the RSF method and (2) to evaluate the changing effects over time of individual pre- and post-treatment variables on predicting mortality. Methods: 1330 hip fracture surgical patients were included. Forty-five admission and in-hospital variables were analyzed as potential predictors of all-cause mortality. A random survival forest (RSF) algorithm was applied in predictors identification. Cox regression models were then constructed. Sensitivity analyses and internal validation were performed to assess the performance of each model. C statistics were calculated and model calibrations were further assessed. Results: Our machine-learning RSF algorithm achieved a c statistic of 0.83 for 30-day prediction and 0.75 for 1-year mortality. Additionally, a COX model was also constructed by using the variables selected by RSF, c statistics were shown as 0.75 and 0.72 when applying in 2-year and 4-year mortality prediction. The presence of post-operative complications remained as the strongest risk factor for both short- and long-term mortality. Variables including fracture location, high serum creatinine, age, hypertension, anemia, ASA, hypoproteinemia, abnormal BUN, and RDW became more important as the length of follow-up increased. Conclusion: The RSF machine-learning algorithm represents a novel approach to identify important risk factors and a risk stratification models for patients undergoing hip fracture surgery is built through this approach to identify those at high risk of long-term mortality.