Chinese Journal of Tissue Engineering Research ›› 2025, Vol. 29 ›› Issue (2): 409-418.doi: 10.12307/2025.234
Previous Articles Next Articles
Wei Mengli1, 2, Zhong Yaping1, 2, Gui Huixian1, Zhou Yiwen1, Guan Yeming1, Yu Shaohua1
Received:
2024-01-15
Accepted:
2024-02-19
Online:
2025-01-18
Published:
2024-05-25
Contact:
Zhong Yaping, PhD, Professor, Doctoral supervisor, Sports Big Data Research Center of Wuhan Sports University, Wuhan 430079, Hubei Province, China; Hubei Sports and Health Innovation and Development Research Center, Wuhan 430079, Hubei Province, China
About author:
Wei Mengli, PhD candidate, Sports Big Data Research Center of Wuhan Sports University, Wuhan 430079, Hubei Province, China; Hubei Sports and Health Innovation and Development Research Center, Wuhan 430079, Hubei Province, China
Supported by:
CLC Number:
Wei Mengli, Zhong Yaping, Gui Huixian, Zhou Yiwen, Guan Yeming, Yu Shaohua. Sports injury prediction model based on machine learning[J]. Chinese Journal of Tissue Engineering Research, 2025, 29(2): 409-418.
Add to citation manager EndNote|Reference Manager|ProCite|BibTeX|RefWorks
提出了功能运动筛查测试,提出了更高效、精准的测评方案。然而,人工测评对诊断医师的临床经验要求高,导致此类测试的稳健度难以保证。HULIN等[13]指出,可以采用统计学手段,基于测试数据构建运动损伤预警模型,从而量化运动损伤风险,以解决人工测评精度易受主观经验影响的问题。基于HULIN等[13]观点,OPAR等[14]采用统计学方法,基于足球运动员的腘绳肌离心力量数据与腘绳肌损伤率建立了线性相关模型。随着可穿戴设备、计算机视觉、精微传感器和GPS定位等先进运动数据采集技术的完善,越来越多与运动损伤风险关联的数据指标被实时跟踪记录,线性模型无法处理多因素间的非线性交互效应,其缺陷亦愈发明显。SHORTLIFFE等[9]指出可加快机器学习等人工智能技术的应用,以快速处理冗余、复杂的数据,提升运动损伤预警模型的速度与精度。POL等[16]论述了基于机器学习技术预测运动损伤风险的基本业务流程、主要业务内容及操作原则。具体时间脉络见图3。各运动损伤预警方法的对比总结见表1。"
以综合研判运动员的损伤风险[10-11]。然而人工判断的精度易受主观经验的影响,且预测反馈滞后,难以满足运动损伤预警需求,因此运动医学界尝试探索可快速量化运动损伤风险的方法。随着循证医学的发展,运动医学界尝试采用统计学手段量化运动员损伤风险,线性分析模型逐步用于揭示各损伤风险因素与运动员损伤率的线性关系,进而统计推断运动员的损伤风险[12]。HULIN等[13]构建了橄榄球运动员训练负荷与运动损伤率的线性回归模型。如OPAR等[14]基于精英级足球运动员的腘绳肌离心力量数据与腘绳肌损伤率建立了线性相关模型。线性模型实现了运动损伤风险的量化评估,有效降低了主观经验错误的风险,提升了运动损伤预警工作的精度与速度。但是,随着可穿戴设备、计算机视觉、精微传感器和GPS定位等先进运动数据采集技术的完善,越来越多与运动损伤风险关联的数据指标被实时跟踪记录,线性模型无法处理多因素间的非线性交互效应的缺陷亦愈发明显[15]。如POL等[16]认为运动损伤本质上是一个复杂的动态事件,该事件的形成受运动员机能状态、训练负荷水平、场地环境等多维因素影响,且各因素间存在复杂的交互作用,线性回归模型难以处理如此复杂的数据关系,因此其预测运动损伤风险的效力有限。 机器学习是一种基于统计学和计算机科学的人工智能技术,通过使用算法和数据构建模型,并自动化推理和预测,从而使机器能够自主学习和优化,进而将人类综合复杂思维能力赋予机器,替代人类执行部分决策活动,以减少决策工作中可能产生的人为错误。机器学习模型注重从大规模数据集中识别出数据运行模式和规律,从而高效处理复杂的多因素交互效应,以预测未来的结果或行为。近年来,运动医学界呼吁采用机器学习技术对高密度、多元异构的运动数据样本进行深度信息挖掘,并分析指标间复杂、非线性的关系,以建立预测精度更高、反馈速度更快的运动损伤智能预警模型,从而替代专家诊断工作[9]。 根据“学习数据”是否拥有标注信息可将机器学习模型分为监督式学习与无监督式学习和半监督式学习3类。所谓“标注”指可直接反映机器学习目标的关键数据指标。例如,运动疲劳程度、运动负荷和比赛天气对运动损伤率的影响,其中的“运动损伤率”即标注。监督学习模型要求学习样本包含完整的标注信息,从给定的训练数据集中学习出一个函数,再将该函数运用于新的数据集来预测结果。无监督学习模型则无需学习样本标注信息,侧重于在大量数据中寻找潜在的规律,将相似的事物归为一类。半监督学习模型是介于监督式与无监督式中的一种机器学习方式,可利用少量的标注数据和大量的未标注数据进行训练和分类[17]。 基于机器学习的运动损伤预警模型的构建需完成如下步骤[18]:①模型特征提取:收集与运动损伤风险密切相关的指标数据集(如身体关节的力量与活动度、训练负荷、疲劳程度、损伤史等),并对数据指标进行清理、去噪、归一化、特征纳入与权重等预处理,从而提炼出与预测结果变量相关的特征指标;②模型主体训练:根据预警的损伤类型,与纳入的特征指标类别,选择合适的预警算法,建立模型训练所需的数据集,从而训练模型主体;③模型评估优化:使用测试数据验证模型的性能,以确定模型是否可以有效地预测运动损伤风险,并对模型进行优化,以提高运动损伤预警模型的性能,见图4。 鉴于运动损伤机器学习预警模型的建立需完成模型特征提取、模型主体训练和模型评估优化3个环节,因此文章将从以上3个方面对现有研究成果进行梳理与归纳,总结现有研究在各模型构建环节的共性与问题,为中国运动损伤预警模型的构建工作提供理论指导。 2.2 基于机器学习的运动损伤预警模型的特征提取 2.2.1 运动损伤预警模型特征指标纳入种类 任何事物都具有其独特的特征,这是事物被区分的基本前提。如一个人的身高、体质量和相貌均具有较强的个体特征,这是其区别于他人的前提。人类智能的形成便是在了解事物特征的基础上,识别事物属性的过程。因此,如何让计算机识别未知事物的特征,从而判断事物的属性,是实现机器智能的先决条件。特征提取的首要内容便是选取可代表事物特点的数据指标输入至计算机,随后设计各指标的阈值,当计算机获取了相关指标数据后,便可根据预先设计的阈值,判断事物的属性[19]。以跑步常见的过度使用性损伤为例,造成该损伤的主要相关指标为运动负荷量,那么该指标便是预测跑步过度使用性损伤风险的主要特征指标[17-18]。 运动损伤病因学理论指出,运动损伤的成因主要可分为内部风险因素及外部风险因素两类[20]。其中,内部风险因素指与人体自身训练适应相关的因素,如人体测量信息、运动素质及运动损伤史等,内部风险因素是决定运动员是否具有损伤倾向的关键原因。外部风险因素指与训练环境、训练过程密切相关的因素,如训练场地的照明情况与配备的安全措施、是否佩戴运动装备、天气条件、训练与比赛负荷等,外部风险因素更多担任了一种损伤诱导的角色,只有当运动员内部风险因素作用使其具有损伤倾向时,外部风险因素才可能对其损伤的发生产生诱导作用,使得运动员变成易受外界刺激影响的高损伤风险群体[21-22]。 现有的机器学习模型基本参照了运动损伤病因学理论,纳入了诸多与损伤密切相关的预测特征指标[26-41]。如表2显示,主要纳入的特征指标类别有人体测量信息(年龄、性别、身体形态;n=14,87.5%)、运动素质(柔韧"
性、肌肉力量、平衡能力;n=8,50%)、训练负荷(n=4,25%)、损伤史(n=4,25%)、训练年限(n=4,25%)、睡眠质量(n=1,6.2%)、个体遗传信息(n=1,6.2%)、学业表现(n=1,6.2%)。从总体上看,模型特征指标纳入种类多为内部风险特征指标,外部风险特征指标仅纳入了训练负荷一项,内外部风险指标纳入比例严重失衡。对已纳入的特征指标具体分析发现,在内部风险指标方面,还较缺乏运动恢复类(如营养摄入、睡眠质量、恢复放松手段等)与身体机能类(能量代谢、疲劳程度等)等重要指标的纳入。在外部风险指标方面,缺乏与比赛场景相关的特征指标(如比赛天气、比赛负荷、比赛场地材质、比赛地形等),而上述指标均已被证实与运动损伤风险具有密切联系,相关特征指标的缺失将限制模型预警性能的进一步提升。因此,后期需进一步完善运动损伤机器学习预警模型的纳入特征指标种类,尤其是外部风险类特征指标的纳入,以获取更为完整的运动损伤预警信息。 2.2.2 运动损伤预警模型的特征指标权重 在真实的运动环境中,各类损伤风险因素的影响力并非是固定的,而是会随着时间推移产生动态及非线性变化。例如,运动员们在每天重复的训练中,其暴露的外部风险因素(如比赛地形、天气条件、训练负荷)通常大体相似,但随着运动员生理机能上的适应(如肌肉力量提升、疲劳恢复速度提高),外部风险因素的作用会下降,即产生所谓的风险因素补偿效应[21]。因此在预警运动损伤时,采用线性相关或回归分析来发掘风险因素与运动损伤间的关联,从而评估各风险因素的权重时,必然会受风险因素补偿效应的影响而产生预警误差,因此需以自组织视角审视各风险因素与运动损伤间的潜在联系,分析各因素间的交互效应,以提炼运动损伤风险变化机制,从而实现运动损伤的精准预警[22]。相较于传统统计模型,机器学习模型的特征权重突出高阶非线性的统计分析方法的应用,以分析与预测结果密切相关的特征指标间动态、非线性的交互作用过程,从而挖掘庞杂、多维数据中潜在的数据规律,进而保证运动损伤预警模型的精准性与稳健性。 当前基于机器学习的运动损伤预警模型的特征指标权重方法主要有3大类[23]:一是基于过滤法(filter),即根据各种统计检验中的分数以及相关性评估各特征指标的权重系数,常见方法有基尼系数、相关性分析、卡方分析、符号秩检验和最小二乘法等;二是基于嵌入法(embedded),即在模型训练过程中直接考虑各特征指标的重要性,常见方法有SHAP图、正则化逻辑回归和套索分析等;三是基于包裹法(wrapper),即利用实际的机器学习模型来评估特征子集的效果,从而选择出最佳特征子集,常见方法有遗传算法、正向选择和反向选择等。不同的损伤类型,相应的风险预测的特征指标亦不同,因此研究者们所采取的特征权重方法亦存在差异。基于此,文章系统梳理了相关案例,以试图总结基于机器学习的运动损伤预警模型的特征权重方法特点。现有研究显示,研究者采用最多的特征权重评价方式为正向选择(n=4,25%),其次分别为基尼系数下降幅度(n=1,6.2%)、套索分析(n=1,6.2%)、最小二乘法(n=1,6.2%)、SHAP概要图(n=1,6.2%)、非线性相关分析(n=1,6.2%)、线性相关分析(n=1,6.2%)、Wilcoxon符号秩检验(n=1,6.2%)、卡方检验(n=1,6.2%)、Fishers精确检验(n=1,6.2%)、正则化逻辑回归(n=1,6.2%),见表2。 基于上述总结发现,现有研究多采用过滤法权重运动损伤预警模型纳入的特征指标,占总研究比例的43.4%,其次是嵌入法(24.8%),包裹法占比最低(12.5%)。相较于嵌入法与包裹法,过滤法通常应用在模型主体训练之前,因此在权重特征指标方面具有速度快、可处理高密度数据等应用优势,但不足之处在于,过滤法更注重明确各特征指标与预测变量间的非线性关系,在量化各特征指标间的交互作用的能力较弱[24-25],这一定程度上不利于深度分析运动损伤风险因素补偿效应,将限制运动损伤预警模型性能的进一步提升。嵌入法与包裹法可以很好地兼顾多损伤风险因素间交互作用的分析,更符合运动损伤预警模型的构建需求,因此后期在开展运动损伤预警模型构建研究时,可积极推进嵌入法与包裹法应用于运动损伤预警模型的特征指标权重工作。 2.3 基于机器学习的运动损伤预警模型的主体训练 机器学习模型的主体训练主要涉及训练集规模设置和训练算法选择两方面内容,以上内容关乎模型的准确性和泛化能力。训练集规模越大,模型越能够学习到训练集的特点和规律。当训练集规模不足时,模型容易发生过拟合(overfitting),即模型过于复杂、过度匹配训练集数据,导致在新数据上表现不佳。相反,如果数据集过大,会导致模型泛化能力的下降(即在训练集上表现良好,但在测试集上表现糟糕),因此准备合理的训练集规模尤为重要[42]。此外,算法选择对模型性能至关重要,按照机器学习方式,机器学习算法也分为监督式、无监督式及半监督式3类,不同的算法适用于不同的问题和数据类型,关乎模型的精度与泛化能力间的平衡,需根据实际问题与数据类型合理选择模型训练算法。 在运动损伤预警模型的训练集规模方面,参考现有标准发现[43],在文章纳入的文献中,大样本训练集 (观测记录次数≥1 000)占比25%、中等样本训练集(100≤观测记录次数< 1 000)占比62.5%、小样本训练集(观测记录次数< 100)占比12.5%。在预警模型构建算法选择方面,主要采用的算法有支持向量机(n=7,43.7%)、随机森林(n=6,37.5%)、决策树(n=5,31.2%)、神经网络(n=3,18.7%)、逻辑回归(n=1,6.2%)。监督式算法为运动损伤机器学习预警模型的主流训练算法,占总研究文献的87.5%,无监督式算法占37.5%,半监督式算法占12.5%。基于上述分析认为,运动损伤机器学习预警模型的主体训练工作有如下特点:①现研究多基于中等样本训练集开展模型主体训练,大样本训练或小样本训练较少;②监督式学习算法为主流的模型主体训练算法,其中以支持向量机、决策树算法应用最为常见,见表3。 基于当前研究在模型主体训练方面的共性特点,分析认为现有研究还存在如下不足:①模型主体训练过于倚重监督式学习算法,需强化半监督与无监督式学习算法的应用。监督式学习算法虽然可依据数据的标注信息"
快速提取数据规律,并基于数据规律对新样本数据进行预测,使得训练出的模型具备较高的预测精度与泛化能力。但在实际体育场景中,很难获取具有完整标注信息的高质量数据[44],因此监督式算法的实际应用场景易受限,而半监督式学习与监督式学习对数据标注信息要求低,且可从无标注信息的数据中提取关键预测特征指标,在一定程度上可弥补监督式学习的应用劣势,更贴合实际体育场景的数据采集分析需求,因此后期可进一步突出应用半监督学习与无监督学习算法训练模型主体;②模型主体训练样本规模单一,训练样本多为中等规模,缺乏小样本、大样本模型的训练实践。相对于小样本训练集,中等样本训练集有更多的样本示例,因此可更好地帮助模型把握数据分布特征,避免出现过度拟合的情况。而相对于大样本训练集,中等样本训练集更容易获取,训练出的模型稳定性更佳[45]。因此,在训练模型主体时,中等样本集通常被认为是兼顾模型预测精确性与稳定性的合理选择。但是,采用中等样本开展模型主体训练并不符合所有运动项目的需求,例如对于数据资源较为丰富的项目(如足球和橄榄球),开展大样本训练易提取出更多的预测特征指标,所得模型的泛化能力更强。而对于数据资源匮乏的运动项目,亦可通过小样本学习技术,降低数据的噪声与不一致性,也可取得不错的预测效果。因此,需基于运动项目数据资源的现状,因势利导地设置合适的模型主体训练样本规模。 2.4 基于机器学习的运动损伤预警模型的评估优化 运动损伤预警模型的优化需经历两大环节:一是对模型性能进行验证与评估,从而明确模型存在的问题;二是对模型进行优化操作,即对模型参数、算法、结构进行调整,以增强模型的预警性能。在早先基于统计学习技术的运动损伤预警模型的研究中,研究者多采用留一法(Leave one out crossvalidation,LOOCV)验证模型性能,即将数据集分成训练集和测试集,通过对比测试集的实际值和模型预测值的误差来评估模型的预警能力,通常采用的模型评估指标主要为R2值、P值等,模型优化操作的主要方法为特征优化,即增加或降维纳入的预测特征指标。相较于统计学模型,用于构建机器学习模型的数据样本大部分由机器自动采集并处理,可用于训练的样本量通常较大[46]。因此,相较于传统的统计学模型,机器学习模型验证评估更注重在数据上的通用性和泛化能力,以避免模型对训练数据过拟合的情况。机器学习模型验证评估主要采用交叉验证方法,将数据集分成训练集、验证集和测试集,通过比较验证集的实际预测值和模型预测值的误差来评估模型的泛化能力,其常用的模型评估指标主要有准确度、AUC值、精确度、灵敏度和F1分数等。在模型优化操作方面,机器学习模型除了常规的特征优化外,还采用数据增强、算法置换及超参数调整等操作方法[45-46]。 不同的损伤类型,用于训练预警模型的数据种类亦不同,因此在运动损伤预警模型构建过程中,不同研究者对模型评估与优化的方法亦具有差异,文章试图梳理现有成果,总结当前运动损伤预警模型评估优化工作的共性特点,为中国相关研究提供借鉴。首先,在模型性能评估方面,现有研究主要采用了HoldOut交叉与k-交叉两种验证方式,分别占总研究文献的56.25%、43.75%,最常采用的性能评价指标为AUC(n=12,75%),其次是灵敏度(n=8,50%)、特异度(n=6,37.5%)、F1分数(n=5,31.2%)、准确度(n=3,18.75%)及精确度(n=3,18.75%)。其次,在模型性能方面,AUC值范围(0.76±0.12),灵敏度范围(75.92±11.03)%,特异度(80.03±4.54)%,F1分数值范围(80.60±10.63)%,准确度范围(69.96±13.10)%,精确度范围(70.00±14.71)%。再次,在预警模型的优化操作方面,数据增强占总研究文献的43.75%,特征优化占比43.75%,算法置换占比18.75%,超参数调整占比18.75%,见表4。 基于上述分析,发现如下研究信息:①现有模型的预测性能仍有较大提升空间。如分析结果显示,现有运动损伤预警模型的准确度、精确度均在70%左右,后期可对模型进一步优化,提升模型的预警性能。②在模型优化操作方面,现研究主要通过改善模型训练数据集质量,从而提升模型预警性能,如增加训练样本量、剔除样本缺失值等数据增强方法,从而降低模型过拟合的可能性,或采用特征优化方法,对预测特征指标进行降维、标准化、归一化处理,提升特征指标的预测效度,而对模型训练算法与模型超参数进行优化调整的研究较少。然而,后两种方法在提升模型鲁棒性与泛化能力方面具有重要意义,基于大量的算法置换或超参数调试,将有效降低运动损伤预警模型陷入局部最优解的风险,以维持模型不同损伤场景下的预测精度,因此后期研究需重视采用多样化的模型优化操作,从而全面提升模型性能。"
[1] REYNOLDS P. Agency for healthcare research and quality: extramural support for rehabilitation activities. Am J Phys Med Rehabil. 2005;84(12):1034-1035. [2] YTTERSTAD B. The harstad injury prevention study: the epidemiology of sports injuries. An 8 year study. Br J Sports Med. 1996;30(1):64-68. [3] WHITELEY R. ‘Moneyball’ and time to be honest about preseason screening: it is a sham making no inroads on the 1 billion dollar injury costs in baseball. Br J Sports Med. 2016;50(14):835-836. [4] THORBORG K, KROMMES KK, ESTEVE E, et al. Effect of specific exercise-based football injury prevention programmes on the overall injury rate in football: a systematic review and meta-analysis of the FIFA 11 and 11+ programmes. Br J Sports Med. 2017;51(7):562-571. [5] 钟亚平,刘鹏.基于RBF神经网络算法的田径运动损伤预警模型研究[J].计算机应用与软件,2014,31(6):48-51. [6] ZADEH A, TAYLOR D, BERTSOS M, et al. Predicting sports injuries with wearable technology and data analysis. Inf Syst Front. 2021;23(2):1023-1037. [7] RAJULA HSR, VERLATO G, MANCHIA M, et al. Comparison of conventional statistical methods with machine learning in medicine: diagnosis, drug development, and treatment. Medicina (Kaunas). 2020;56(9):455-465. [8] MAJUMDAR A, BAKIROV R, HODGES D, et al. Machine learning for understanding and predicting injuries in football. Sports Med Open. 2022; 8(1):73-83.
[9] SHORTLIFFE EH, SEPÚLVEDA MJ. Clinical decision support in the era of artificial intelligence. J Am Med Dir Assoc. 2018;320(21):2199-2200. [10] BUTLER DL, NOYES FR, GROOD ES. Ligamentous restraints to anterior-posterior drawer in the human knee. A biomechanical study. J Bone Joint Surg Am. 1980;62(2):259-270. [11] COOK G, BURTON L, HOOGENBOOM BJ, et al. Functional movement screening: the use of fundamental movements as an assessment of function-part 1. Int J Sports Phys Ther. 2014;9(3):396-409. [12] ADAMHULME, FINCH C. From monocausality to systems thinking: a complementary and alternative conceptual approach for better understanding the development and prevention of sports injury. Int J Epidemiol. 2015;2(1):1-12. [13] HULIN BT , GABBETT TJ , BLANCH P, et al. Spikes in acute workload are associated with increased injury risk in elite cricket fast bowlers. Br J Sports Med. 2014;48(8):708-712. [14] OPAR DA, WILLIAMS MD, TIMMINS RG, et al. Eccentric hamstring strength and hamstring injury risk in Australian footballers. Med Sci Sports Exerc. 2015;47(4):857-865. [15] CLAUDINO JG, CAPANEMA DO, SOUZA TV, et al. Current approaches to the use of artificial intelligence for injury risk assessment and performance prediction in team sports: a systematic review. Sports Med Open. 2019;5(1):1-12. [16] POL R, HRISTOVSKI R, MEDINA D, et al. From microscopic to macroscopic sports injuries. Applying the complex dynamic systems approach to sports medicine: a narrative review. Br J Sports Med. 2019;53(19):1214-1220. [17] SARKER IH. Machine learning: algorithms, real-world applications and research directions. SN Comput Sci. 2021;2(3):1-21. [18] VAN EETVELDE H, MENDONCA LD, LEY C, et al. Machine learning methods in sport injury prediction and prevention: a systematic review. J Exp Orthop. 2021;8(1):27-42. [19] KOKKOTIS C, MOUSTAKIDIS S, PAPAGEORGIOU E, et al. Machine learning in knee osteoarthritis: a review. Osteoarthr Cartil Open. 2020;2(3):100069. [20] BAHR R. Understanding injury mechanisms a key component of preventing injury in sport. Br J Sports Med. 2005;39(6):324-329. [21] MEEUWISSE WH, TYREMAN H, HAGEL B, et al. A dynamic model of etiology in sport injury: the recursive nature of risk and causation. Clin J Sport Med. 2007;17(3):215-219. [22] WINDT J, GABBETT TJ. How do training and competition workloads relate to injury? The workload—injury aetiology model. Br J Sports Med. 2016;51(5): 428-435. [23] BITTENCOURT NFN, MEEUWISSE WH, MENDON ALD, et al. Complex systems approach for sports injuries: moving from risk factor identification to injury pattern recognition—narrative review and new concept. Br J Sports Med. 2016;50(21):1309-1314. [24] PUDJIHARTONO N, FADASON T, KEMPALIEHR AW, et al. A review of feature selection methods for machine learning-based disease risk prediction. Front Bioinform. 2022;27(2):1-17. [25] SAROJ A, MONALI B, BISWAJIT P. Review on feature selection and classification using neuro-fuzzy approaches. Int J Appli Evolu Comput. 2016;7(4):28-44. [26] MCCULLAGH J, WHITFORT T. An investigation into the application of artificial neural networks to the prediction of injuries in sportInt. J Sport Health Sci. 2013;7(7):1356-1360. [27] WHITESIDE D, MARTINI DN, LEPLEY AS, et al. Predictors of ulnar collateral ligament reconstruction in major league baseball pitchers. Am J Sports Med. 2016;3(1):1-8. [28] CAREY DL, CROSSLEY KM, WHITELEY R, et al. Modeling training loads and injuries: the dangers of discretization. Med Sci Sports Exerc. 2018;50(11):1-8. [29] LÓPEZ V, ALEJANDRO, AYALA F, et al. Preventive model for muscle injuries: a novel approach based on learning algorithms. Med Sci Sports Exerc. 2018;50(5):915-927. [30] ALESSIO R, LUCA P, PAOLO C, et al. Effective injury forecasting in soccer with GPS training data and machine learning. Plos One. 2018;13(7):1-15. [31] RUDDY JD, SHIELD AJ, MANIAR N, et al. Predictive modeling of hamstring strain injuries in elite australian footballers. Med Sci Sports Exerc. 2018;50(5):906-914. [32] AYALA, FRANCISCOLOPEZ V, ALEJANDROGAMEZ M, et al. A preventive model for hamstring injuries in professional soccer: learning algorithms. Int J Sports Med. 2019;40(5):1-27. [33] RODAS G, OSABA L, ARTETA D, et al. Genomic prediction of tendinopathy risk in elite team sports. Int J Sports Physiol Perform. 2019;15(4):1-7. [34] LUU BC, WRIGHT AL, HAEBERLE HS, et al. Machine learning outperforms logistic regression analysis to predict next-season nhl player injury: an analysis of 2322 players from 2007 to 2017. Orthop J Sports Med. 2020; 8(9):1-8. [35] OLIVER B, FA C, MDSC D, et al. Using machine learning to improve our understanding of injury risk and prediction in elite male youth football players. J Sci Med Sport. 2020;23(11):1044-1048. [36] ROMMERS N, RSSLER R, VERHAGEN E, et al. A machine learning approach to assess injury risk in elite youth football players. Med Sci Sports Exerc. 2020;52(8):1745-1751. [37] CASTELLANOS J, PHOO CP, ECKNER JT, et al. Predicting risk of sport-related concussion in collegiate athletes and military cadets: a machine learning approach using baseline data from the care consortium study. Sports Med. 2020;51(3):567-579. [38] TABORRI J, MOLINARO L, SANTOSPAGNUOLO A, et al. A machine-learning approach to measure the anterior cruciate ligament injury risk in female basketball players. Sensors. 2021;21(9):3141-3158. [39] RAHLF AL, HOENIG T, STÜRZNICKEL J, et al. A machine learning approach to identify risk factors for running-related injuries: study protocol for a prospective longitudinal cohort trial. BMC Sports Sci Med Rehabil. 2022; 14(1):1-11. [40] JAUHIAINEN S, KAUPPI JP, KROSSHAUG T, et al. Predicting acl injury using machine learning on data from an extensive screening test battery of 880 female elite athletes. Am J Sports Med. 2022;5(11):2917-2924. [41] BOGAERT S, DAVIS J, VAN RS. Impact of gender and feature set on machine-learning-based prediction of lower-limb overuse injuries using a single trunk-mounted accelerometerhe. Sensor. 2022;22(11):2860-2877. [42] Charilaou P, Battat R. Machine learning models and over-fitting considerations. World J Gastroenterol. 2022;28(5):605-607. [43] RAJPUT D, WANG WJ, CHEN CC. Evaluation of a decided sample size in machine learning applications. BMC Bioinformatics. 2023;24(1):48-65. [44] VONDRICK C , PATTERSON D , RAMANAN D. Efficiently scaling up crowdsourced video annotation: a set of best practices for high quality, economical video labeling. Int J Comput Vis. 2013;101(1):184-204. [45] VABALAS A, GOWEN E, POLIAKOFF E. Machine learning algorithm validation with a limited sample size. PLoS One. 2019;14(11):1-20. [46] DEIST TM, PATTI A, WANG Z, et al. Simulation-assisted machine learning. Bioinformatics. 2019;35(20):4072-4080. [47] GABBETT TJ. The training-injury prevention paradox: should athletes be training smarter and harder? Br J Sports Med. 2016;50(5):273-280. [48] CHAMBERS R, GABBETT TJ, COLE MH, et al. The use of wearable microsensors to quantify sport-specific movements. Sports Med. 2015; 45(7):1065-1081. [49] WEHBE GM, GABBETT TJ, HARTWIG TB, et al. Reliability of a cycle ergometer peak power test in running-based team sport athletes: a technical report. J Strength Cond Res. 2015;29(7):2050-2055. [50] GALLO T, CORMACK S, GABBETT T, et al. Characteristics impacting on session rating of perceived exertion training load in Australian footballers. J Sports Sci. 2015;33(5):467-475. [51] KALKHOVEN JT, WATSFORD ML, IMPELLIZZERI FM. A conceptual model and detailed framework for stress-related, strain-related, and overuse athletic injury. J Sci Med Sport. 2020;23(8):726-734. [52] TOOHEY LA, DREW MK, FORTINGTON LV, et al. An updated subsequent injury categorisation model (SIC-2.0): data-driven categorisation of subsequent injuries in sport. Sports Med. 2018;48(9):2199-2210. [53] SHRIER I, STEELE RJ, ZHAO M, et al. A multistate framework for the analysis of subsequent injury in sport (M-FASIS). Scand J Med Sci Sports. 2016; 26(2):128-139. [54] SHRIER I, STEELE RJ. Classification systems for reinjuries: a continuing challenge. Br J Sports Med. 2014;48(18):1338-1339. [55] KIEFER AW, KUSHNER AM, GROENE J, et al. A Commentary on real-time biofeedback to augment neuromuscular training for ACL injury prevention in adolescent athletes. J Sports Sci Med. 2015;14(1):1-8. [56] BIRÓ A, CUESTA-VARGAS AI, SZILÁGYI L. AI-assisted fatigue and stamina control for performance sports on IMU-generated multivariate times series datasets. Sensors (Basel). 2023;24(1):132-139. [57] WILKERSON GB, GUPTA A, COLSTON MA. Mitigating sports injury risks using internet of things and analytics approaches. Risk Anal. 2018;38(7):1348-1360. [58] HULIN BT, GABBETT TJ, JOHNSTON RD, et al. Wearable microtechnology can accurately identify collision events during professional rugby league match-play. J Sci Med Sport. 2017;20(7):638-642. [59] CUI J, DU H, WU X. Data analysis of physical recovery and injury prevention in sports teaching based on wearable devices. Prev Med. 2023;173:107589. [60] JOHNSON WR, MIAN A, LLOYD DG, et al. On-field player workload exposure and knee injury risk monitoring via deep learning. J Biomech. 2019;93:185-193. [61] CHAWLA S, ESTEPHAN J, GUDMUNDSSON J, et al. Classification of passes in football matches using spatiotemporal data. ACM Trans Spat Algorithms Syst. 2017;3(2):1-30. |
[1] | Li Jiagen, Chen Yueping, Huang Keqi, Chen Shangtong, Huang Chuanhong. The construction and validation of a prediction model based on multiple machine learning algorithms and the immunomodulatory analysis of rheumatoid arthritis from the perspective of mitophagy [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(在线): 1-15. |
[2] | Wang Mi, Ma Shujie, Liu Yang, Qi Rui. Identification and validation of characterized gene NFE2L2 for ferroptosis in ischemic stroke [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(7): 1466-1474. |
[3] | Liu Yani, Yang Jinghuan, Lu Huihui, Yi Yufang, Li Zhixiang, Ou Yangfu, Wu Jingli, Wei Bing . Screening of biomarkers for fibromyalgia syndrome and analysis of immune infiltration [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(5): 1091-1100. |
[4] | Yang Bin, Tao Guangyi, Yang Shun, Xu Junjie, Huang Junqing . Visualization analysis of research hotspots of artificial intelligence in field of spinal cord nerve injury and repair [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(4): 761-770. |
[5] | Yang Dingyan, Yu Zhenqiu, Yang Zhongyu. Machine learning-based analysis of neutrophil-associated potential biomarkers for acute myocardial infarction [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(36): 7909-7920. |
[6] | Fang Ying, Zhang Yanwei, Li Xi, Yan Peidong, Bi Miao. Improvements in automatic diagnosis methods for knee osteoarthritis based on deep learning [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(35): 7511-7518. |
[7] | Jiang Qiyu, Zeng Huiyan. A novel analysis and prediction method for potential mechanisms of traditional Chinese medicine based on artificial intelligence and omics data-driven approach [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(35): 7552-7561. |
[8] | Chen Ying, Guo Xiaojing, Mo Xueni, Ma Wei, Wu Shangzhi, Li Xiangling, Xie Tingting. Analysis of diagnostic biomarkers for ischemic stroke and experimental validation of targeted cuproptosis related genes [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(35): 7562-7570. |
[9] | Gong Yuehong, Wang Mengjun, Ren Hang, Zheng Hui, Sun Jiajia, Liu Junpeng, Zhang Fei, Yang Jianhua, Hu Junping. Machine learning combined with bioinformatics screening of key genes for pulmonary fibrosis associated with cellular autophagy and experimental validation [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(35): 7679-7689. |
[10] | Li Jing, Lu Guangqi, Zhuang Minghui, Cui Ying, Yu Zhangjingze, Sun Xinyue, Ma Mingming, Zhu Liguo, Yu Jie. Development of a clinical prediction model for cervical instability in young and middle-aged adults based on machine learning [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(33): 7203-7210. |
[11] | Wang Simin, Zhang Dezhou, Zhao Jing, Wang Chaoqun, Li Kun, Chen Jie, Bai Xue, Zhao Hailong, Zhang Shaojie, Ma Yuan, Hao Yunteng, Yang Yang, Li Zhijun, Shi Jun, Wang Xing. Artificial intelligence and cervical spine image recognition: application prospects and challenges [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(33): 7231-7240. |
[12] | Tian Yushi, Fu Qiang, Li Ji . Bioinformatics identification and validation of mitochondrial genes related to acute myocardial infarction [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(31): 6697-6707. |
[13] | Song Haoran, Zhang Yuqiang, Gu Na, Zhi Xiaodong, Wang Wei. Visualization analysis of artificial intelligence in bone trauma research based on Citespace [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(3): 493-502. |
[14] | Zhang Ziyu, Chen Longhao, Sheng Wei, Lyu Hanzhe, Shen Ying, Wang Binghao, Lyu Zhizhen, Lyu Lijiang. Application of artificial intelligence in the diagnosis and treatment of lumbar disc herniation: evolution towards standardization, efficiency, and precision of diagnosis and treatment methods [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(29): 6269-6276. |
[15] | Wang Jiaru, Zhang Ying, Yang Yong, Qi Wen, Xiao Huaye, Ma Qiuping, Yang Lianzhao, Luo Ziwei, He Yaqing, Zhang Jiangyin, Wei Jiawen, Meng Yuan, Tan Silian. Systematic review of machine learning models for predicting functional recovery and prognosis in stroke [J]. Chinese Journal of Tissue Engineering Research, 2025, 29(29): 6317-6325. |
Viewed | ||||||
Full text |
|
|||||
Abstract |
|
|||||