Department of Medical and Surgical Sciences, The Magna Græcia University of Catanzaro, Catanzaro, Italy -
MusculoSkeletalHealth@UMG, The Magna Græcia University of Catanzaro, Catanzaro, Italy -
J Sports Med Phys Fitness. 2024 Mar;64(3):293-300. doi: 10.23736/S0022-4707.23.15417-X. Epub 2023 Dec 21.
Impaired physical performance and muscle strength are recognized risk factors for fragility fractures, frequently associated with osteoporosis and sarcopenia. However, the integration of muscle strength and physical performance in the comprehensive assessment of fracture risk is still debated. Therefore, this cross-sectional study aimed to assess the potential role of hand grip strength (HGS) and short physical performance battery (SPPB) for predicting fragility fractures and their correlation with Fracture Risk Assessment Tool (FRAX) with a machine learning approach.
In this cross-sectional study, a group of postmenopausal women underwent assessment of their strength, with the outcome measured using the HSG, their physical performance evaluated using the SPPB, and the predictive algorithm for fragility fractures known as FRAX. The statistical analysis included correlation analysis using Pearson's r and a decision tree model to compare different variables and their relationship with the FRAX Index. This machine learning approach allowed to create a visual decision boundaries plot, providing a dynamic representation of variables interactions in predicting fracture risk.
Thirty-four patients (mean age 63.8±10.7 years) were included. Both HGS and SPPB negatively correlate with FRAX major (r=-0.381, P=0.034; and r=-0.407, P=0.023 respectively), whereas only SPPB significantly correlated with an inverse proportionality to FRAX hip (r=-0.492, P=0.001). According to a machine learning approach, FRAX major ≥20 and/or hip ≥3 might be reported for an SPPB<6. Concurrently, HGS<17.5 kg correlated with FRAX major ≥20 and/or hip ≥3.
In light of the major findings, this cross-sectional study using a machine learning model related SPPB and HGS to FRAX. Therefore, a precise assessment including muscle strength and physical performance might be considered in the multidisciplinary assessment of fracture risk in post-menopausal women.
身体机能下降和肌肉力量减弱是脆性骨折的公认风险因素，常与骨质疏松症和肌肉减少症相关。然而，肌肉力量和身体机能在综合骨折风险评估中的整合仍存在争议。因此，这项横断面研究旨在评估握力（HGS）和简易体能测试（SPPB）对预测脆性骨折的潜在作用，并采用机器学习方法研究其与骨折风险评估工具（FRAX）的相关性。
在这项横断面研究中，一组绝经后女性接受了力量评估，使用 HGS 测量其握力，使用 SPPB 评估其身体机能，并使用已知的脆性骨折预测算法 FRAX。统计分析包括使用 Pearson's r 进行相关性分析和决策树模型，以比较不同变量及其与 FRAX 指数的关系。这种机器学习方法可以创建一个可视化决策边界图，提供变量之间相互作用预测骨折风险的动态表示。
共纳入 34 例患者（平均年龄 63.8±10.7 岁）。HGS 和 SPPB 均与 FRAX 主要骨折（r=-0.381，P=0.034；r=-0.407，P=0.023）呈负相关，而仅 SPPB 与 FRAX 髋部骨折（r=-0.492，P=0.001）呈负相关。根据机器学习方法，如果 SPPB<6，则 FRAX 主要骨折≥20 和/或髋部骨折≥3 的风险较高。同时，如果 HGS<17.5kg，则 FRAX 主要骨折≥20 和/或髋部骨折≥3 的风险较高。
基于主要发现，本横断面研究使用机器学习模型将 SPPB 和 HGS 与 FRAX 相关联。因此，在绝经后妇女的骨折风险多学科评估中，可能需要考虑对肌肉力量和身体机能进行精确评估。
