Predict Injury Risk Research Articles

Using Interpretable Machine Learning to Predict Injury Risk Among Collegiate Male Basketball Players

Using Interpretable Machine Learning to Predict Injury Risk Among Collegiate Male Basketball Players

Multiscale Structure of Brain and Challenges in Traumatic Brain Injury Risk Prediction

Multiscale Structure of Brain and Challenges in Traumatic Brain Injury Risk Prediction

Explainable Artificial Intelligence for Early Prediction of Pressure Injury Risk.

Hospital-acquired pressure injuries (HAPIs) have a major impact on patient outcomes in intensive care units (ICUs). Effective prevention relies on early and accurate risk assessment. Traditional risk-assessment tools, such as the Braden Scale, often fail to capture ICU-specific factors, limiting their predictive accuracy. Although artificial intelligence models offer improved accuracy, their "black box" nature poses a barrier to clinical adoption. To develop an artificial intelligence-based HAPI risk-assessment model enhanced with an explainable artificial intelligence dashboard to improve interpretability at both the global and individual patient levels. An explainable artificial intelligence approach was used to analyze ICU patient data from the Medical Information Mart for Intensive Care. Predictor variables were restricted to the first 48 hours after ICU admission. Various machine-learning algorithms were evaluated, culminating in an ensemble "super learner" model. The model's performance was quantified using the area under the receiver operating characteristic curve through 5-fold cross-validation. An explainer dashboard was developed (using synthetic data for patient privacy), featuring interactive visualizations for in-depth model interpretation at the global and local levels. The final sample comprised 28 395 patients with a 4.9% incidence of HAPIs. The ensemble super learner model performed well (area under curve = 0.80). The explainer dashboard provided global and patient-level interactive visualizations of model predictions, showing each variable's influence on the risk-assessment outcome. The model and its dashboard provide clinicians with a transparent, interpretable artificial intelligence-based risk-assessment system for HAPIs that may enable more effective and timely preventive interventions.

Predictors of Anterior Cruciate Ligament Tears in Adolescents and Young Adults

Background: Anterior cruciate ligament (ACL) injury in youth can lead to expensive treatment, lengthy rehabilitation, and long-term impairment. Injuries to the ACL are more common in adolescents who participate in organized sports. Purpose: To examine whether there is an association between age, physical activity patterns, pubertal timing (Tanner stage), or body mass index (obesity/overweight status) and the risk of developing an ACL injury in youth. Study Design: Cohort study; Level of evidence, 2. Methods: Data from the Growing Up Today Study (GUTS), a prospective cohort study of youth throughout the United States, were used in this study. From 1996 to 2003, questionnaires were sent to GUTS participants every 12 to 18 months to assess a variety of self-reported factors; in a related 2004 questionnaire, ACL tears and other diagnoses among the participants were reported. Cox proportional hazard models were run to determine the sex-specific association of activity patterns and pubertal timing with risk of developing an ACL tear. Results: A total of 4519 boys and 5622 girls (age: 9-15 years) were included, with 2.4% and 2.0% sustaining an ACL injury, respectively. Age was found to be a significant risk factor for ACL injury in both sexes (male: hazard ratio [HR] = 1.30 [95% CI, 1.14-1.48]; female: HR = 1.23 [95% CI, 1.10-1.38]). Mean hours per week engaged in vigorous activity was also predictive of a higher risk of ACL injury (male: HR = 1.04 [95% CI, 1.01-1.08]; female: HR = 1.10 [95% CI, 1.07-1.14]). The sports that were the most predictive of injury were running, basketball, soccer, and cheerleading/gymnastics in girls and running, football, and karate/martial arts in boys. Additionally, in boys, overweight/obese status was found to be a significant predictor of ACL injury (HR = 2.61 [95% CI, 1.60-4.26]). Furthermore, among 9- to 13-year-old girls, advanced Tanner stage was a strong predictor of injury risk (HR = 2.43 [95% CI, 1.10-5.36]). Conclusion: Age, time engaged in vigorous activity, overweight/obese status, and advanced Tanner stage were associated with an increased risk of ACL injury in young adults and adolescents. Further research with follow-up data will help complement this study and continue to highlight the risk factors associated with ACL reconstruction in adolescents and young adults.

Limitations of Separating Athletes into High or Low-Risk Groups based on a Cut-Off. A Clinical Commentary.

Athlete injury risk assessment and management is an important, yet challenging task for sport and exercise medicine professionals. A common approach to injury risk screening is to stratify athletes into risk groups based on their performance on a test relative to a cut-off threshold. However, one potential reason for ineffective injury prevention efforts is the over-reliance on identifying these 'at-risk' groups using arbitrary cut-offs for these tests and measures. The purpose of this commentary is to discuss the conceptual and technical issues related to the use of a cut-off in both research and clinical practice. How can we better assess and interpret clinical tests or measures to enable a more effective injury risk assessment in athletes? Cut-offs typically lack strong biologic plausibility to support them; and are typically derived in a data-driven manner and thus not generalizable to other samples. When a cut-off is used in analyses, information is lost, leading to potentially misleading results and less accurate injury risk prediction. Dichotomizing a continuous variable using a cut-off should be avoided. Using continuous variables on its original scale is advantageous because information is not discarded, outcome prediction accuracy is not lost, and personalized medicine can be facilitated. Researchers and clinicians are encouraged to analyze and interpret the results of tests and measures using continuous variables and avoid relying on singular cut-offs to guide decisions. Injury risk can be predicted more accurately when using continuous variables in their natural form. A more accurate risk prediction will facilitate personalized approaches to injury risk mitigation and may lead to a decline in injury rates. 5.

"Effect of Progressive Structured Neuromuscular Training Program on Balance, Performance and Injury Risk Prediction in Amateur Futsal Team Players: Protocol for a Randomized Controlled Trial"

"Neuromuscular training improves neuromuscular control and balance, consequently enhancing joint stability. This study aims to assess the impact of a structured neuromuscular training program on balance and performance among university amateur male futsal team players and to analyze biomechanical factors pertinent to injury risk assessments. Thirty athletes aged between 18 and 25, participating in the university male futsal team, will be enrolled in this study. At the outset, all athletes will undergo evaluation using the Korebalance® System, Countermovement Jump, Running-Based Anaerobic Sprint Test, Shot on Goal Test, Sit and Reach Test, Shuttle, Y Balance-Lower Quadrant, and The Tuck Jump test. Subsequently, they will maintain their regular training regimen of classical training twice a week for eight weeks and undergo reassessment using the same battery of tests after 8 weeks of period. Following this initial phase, the athletes will be randomly assigned to two groups: Control Group (n = 15) and Neuromuscular Training Group (n = 15). The Control Group (CG) will continue with their classical futsal training twice a week for eight weeks, while the Neuromuscular Training Group (NMEG) will engage in progressive structured neuromuscular training in addition to their regular futsal training, also twice a week for eight weeks. At the conclusion of the eight-week intervention period, both groups will undergo reevaluation using the same assessment methods. The study commenced in October 2023 and is anticipated to conclude by August 2024. It is anticipated that the integration of progressive structured neuromuscular training into futsal team training programs will yield positive outcomes."

Acute brain injury risk prediction models in venoarterial extracorporeal membrane oxygenation patients with tree-based machine learning: An Extracorporeal Life Support Organization Registry analysis

ObjectiveWe aimed to determine if machine learning can predict acute brain injury and to identify modifiable risk factors for acute brain injury in patients receiving venoarterial extracorporeal membrane oxygenation. MethodsWe included adults (age ≥18 years) receiving venoarterial extracorporeal membrane oxygenation or extracorporeal cardiopulmonary resuscitation in the Extracorporeal Life Support Organization Registry (2009-2021). Our primary outcome was acute brain injury: central nervous system ischemia, intracranial hemorrhage, brain death, and seizures. We used Random Forest, CatBoost, LightGBM, and XGBoost machine learning algorithms (10-fold leave-1-out cross-validation) to predict and identify features most important for acute brain injury. We extracted 65 total features: demographics, pre-extracorporeal membrane oxygenation/on-extracorporeal membrane oxygenation laboratory values, and pre-extracorporeal membrane oxygenation/on-extracorporeal membrane oxygenation settings. ResultsOf 35,855 patients receiving venoarterial extracorporeal membrane oxygenation (nonextracorporeal cardiopulmonary resuscitation) (median age of 57.8 years, 66% were male), 7.7% (n = 2769) experienced acute brain injury. In venoarterial extracorporeal membrane oxygenation (nonextracorporeal cardiopulmonary resuscitation), the area under the receiver operator characteristic curves to predict acute brain injury, central nervous system ischemia, and intracranial hemorrhage were 0.67, 0.67, and 0.62, respectively. The true-positive, true-negative, false-positive, false-negative, positive, and negative predictive values were 33%, 88%, 12%, 67%, 18%, and 94%, respectively, for acute brain injury. Longer extracorporeal membrane oxygenation duration, higher 24-hour extracorporeal membrane oxygenation pump flow, and higher on-extracorporeal membrane oxygenation partial pressure of oxygen were associated with acute brain injury. Of 10,775 patients receiving extracorporeal cardiopulmonary resuscitation (median age of 57.1 years, 68% were male), 16.5% (n = 1787) experienced acute brain injury. The area under the receiver operator characteristic curves for acute brain injury, central nervous system ischemia, and intracranial hemorrhage were 0.72, 0.73, and 0.69, respectively. Longer extracorporeal membrane oxygenation duration, older age, and higher 24-hour extracorporeal membrane oxygenation pump flow were associated with acute brain injury. ConclusionsIn the largest study predicting neurological complications with machine learning in extracorporeal membrane oxygenation, longer extracorporeal membrane oxygenation duration and higher 24-hour pump flow were associated with acute brain injury in nonextracorporeal cardiopulmonary resuscitation and extracorporeal cardiopulmonary resuscitation venoarterial extracorporeal membrane oxygenation.

Injury Risk Predictions in Lunar Terrain Vehicle (LTV) Extravehicular Activities (EVAs): A Pilot Study

Extravehicular activities will play a crucial role in lunar exploration on upcoming Artemis missions and may involve astronauts operating a lunar terrain vehicle (LTV) in a standing posture. This study assessed kinematic response and injury risks using an active muscle human body model (HBM) restrained in an upright posture on the LTV by simulating dynamic acceleration pulses related to lunar surface irregularities. Linear accelerations and rotational displacements of 5 lunar obstacles (3 craters; 2 rocks) over 5 slope inclinations were applied across 25 simulations. All body injury metrics were below NASA’s injury tolerance limits, but compressive forces were highest in the lumbar (250–550N lumbar, tolerance: 5300N) and lower extremity (190–700N tibia, tolerance: 1350N) regions. There was a strong association between the magnitudes of body injury metrics and LTV resultant linear acceleration (ρ = 0.70–0.81). There was substantial upper body motion, with maximum forward excursion reaching 375 mm for the head and 260 mm for the chest. Our findings suggest driving a lunar rover in an upright posture for these scenarios is a low severity impact presenting low body injury risks. Injury metrics increased along the load path, from the lower body (highest metrics) to the upper body (lowest metrics). While upper body injury metrics were low, increased body motion could potentially pose a risk of injury from flail and occupant interaction with the surrounding vehicle, suit, and restraint hardware.

The Injury Risk Prediction of Firefighters with Biomechanical Parameters during Single- and Double-Leg Jumps

The incidence of knee injuries during firefighter training is high, but there is a lack of research on predictive factors and risk assessment for such injuries. Biomechanical assessments can provide a better understanding of how the body’s load changes during exercise, which may alter the risk of injury. Ten firefighters were recruited for our study. Each participant completed the FMS test and the single- and double-leg jump tasks. Motion information was collected and musculoskeletal models of the participants were constructed using OpenSim 4.4 to obtain the joint angle and joint moment. The peak GRF and CoM-CoP angle were also calculated. The findings showed a significantly larger PKFM (p = 0.0195), VAFM (p = 0.0039), and peak AP GRF (p = 0.0039) during the single-leg jump. The opposite performance was observed for KFA (p = 0.0098) and MPA (p = 0.0273). A stepwise multiple linear regression analysis was used to explore the relationship between these parameters and both the FMS score and the times of injuries. The risk of injury is higher in the single-leg jump compared to the double-leg jump. The biomechanical parameters of these two jumps can be used to assess sports injuries and to provide methods and references for injury risk monitoring during firefighter occupational training.

Functional movement screen score to predict injury risk of sports students: a review of foot shape and body mass index

Background and Study Aim. Excessive physical activity poses a risk of injury, particularly when coupled with high exercise intensity. Factors such as Body Mass Index (BMI) and foot morphology may contribute to this risk. This study seeks to evaluate the relationship between BMI, foot shape, and the likelihood of injury among sports students. Material and Methods. A total of 119 sports students, comprising 33 females and 86 males, participated in this study. Purposive sampling was employed to select participants. BMI was determined using height and weight measurements, while foot shape was assessed through footprint measurements. The Functional Movement Screening (FMS) instrument, encompassing 7 movements: Deep Squats, Hurdle Steps, Inline Lunges, Shoulder Mobility, Active Straight Leg Raises, Trunk Stability Push-ups, and Rotary Stability, was utilized to assess injury risk. Data analysis was conducted using Microsoft Excel and IBM SPSS v.25.0, employing descriptive statistics. Results. Based on data analysis, the FMS value in females shows a mean of 18.7 and a standard deviation of 1.21. At the same time, the FMS value in males shows 18.5 and a standard deviation of 1.41. BMI values in females showed a mean of 21.6 and a standard deviation of 3.03. At the same time, BMI values in males showed a mean of 22.9 and a standard deviation of 4.62. The results were that all FMS scores were ≥14. While of BMI measurements, some samples obtained ˂18 kg/m² and ˃30kg/m². BMI measurements according to gender were mainly in the normal weight category. However, those who fall into the obesity category are primarily males. With the results of measuring the foot's shape, the percentage of flat feet is only 0.84% in females and 7.59 in males, while the rest of the sample has a normal foot shape. Conclusions. Individuals with a BMI in the overweight to obese category are more susceptible to high-category injuries, likely due to limited motion range. However, the study did not find evidence linking flat foot shape to high-category injury risk. This inability to establish a correlation may be attributed to the predominantly normal foot shape observed in the sample, resulting in a prediction of low-category injury risk.

Injury Risk Prediction and Prevention Algorithm for Athletes Based on Data Mining

The injury risk prediction and prevention algorithm for athletes employs data mining techniques to analyze various factors contributing to athlete injuries, such as training load, biomechanics, previous injuries, and environmental conditions. By collecting and analyzing vast amounts of data from wearable devices, training logs, and medical records, the algorithm identifies patterns and trends associated with injury occurrence. Using machine learning models, it predicts the likelihood of an athlete sustaining an injury based on these patterns. Additionally, the algorithm provides personalized injury prevention strategies tailored to each athlete's risk profile, including adjusting training regimens, modifying technique, and recommending recovery protocols. Injury prevention is a critical aspect of athlete safety and performance optimization in sports. This paper proposed a novel PMin-Max injury risk prediction and prevention algorithm designed to provide comprehensive assessments of injury risks for athletes. The algorithm leverages data mining techniques to compute both best-case and worst-case probabilities of injury, enabling practitioners to understand the range of potential outcomes and associated uncertainties. By computing minimum and maximum probabilities of injury for individual athletes and generating risk scores, the algorithm facilitates tailored preventive measures based on the severity of predicted injury risks. Through simulation experiments and analysis of simulated results, we demonstrate the algorithm's effectiveness in identifying athletes with elevated injury risks and highlighting situations where injury risks are relatively low.

Identification of Potential Injury Risk Factors and Prediction Model Construction of Athletes using Data Mining Algorithm

Identification of potential injury risk factors and construction of prediction models for athletes using data mining algorithms involve analyzing various factors such as biomechanical, physiological, and environmental variables to determine their correlation with injury occurrences. Data mining techniques, such as decision trees, logistic regression, or neural networks, are then applied to identify patterns and relationships within the data. By integrating this information, predictive models can be developed to forecast the likelihood of athletes sustaining injuries based on their individual characteristics and training conditions. This paper introduces a novel approach for athlete injury risk prediction using the Noninvasive Data Mining Model with Multi-relational and Multi-dimensional Clustering (Ni-DMMMC) algorithm. By leveraging advanced data mining techniques, Ni-DMMMC aims to identify hidden relationships between various athlete characteristics, including biomechanical risk, training load, physiological markers, and environmental factors, to predict injury susceptibility accurately. Through multi-relational and multi-dimensional clustering, the algorithm effectively categorizes athletes into distinct clusters based on their unique risk profiles. These clusters provide valuable insights into the diverse injury risk profiles present within athlete populations, enabling targeted intervention strategies and personalized athlete management approaches. The proposed approach holds significant promise for enhancing injury prevention efforts, optimizing athlete performance, and improving overall well-being in sports medicine practice. athletes in Cluster 1 exhibit high biomechanical risk (0.85), moderate training load (0.60), low physiological markers (0.25), and moderate environmental risk (0.45), while athletes in Cluster 2 display moderate biomechanical risk (0.70), high training load (0.90), high physiological markers (0.80), and low environmental risk (0.30). These numerical values serve as indicators for assessing an athlete's injury risk profile comprehensively.

Development and validation of a risk prediction model for community-acquired pressure injury in a cancer population: A case-control study

BackgroundPatients with cancer are susceptible to pressure injuries, which accelerate deterioration and death. In patients with post-acute cancer, the risk of pressure injury is ignored in home or community settings. ObjectiveTo develop and validate a community-acquired pressure injury risk prediction model for cancer patients. MethodsAll research data were extracted from the hospital's electronic medical record system. The identification of optimal predictors is based on least absolute shrinkage and selection operator regression analysis combined with clinical judgment. The performance of the model was evaluated by drawing a receiver operating characteristic curve and calculating the area under the curve (AUC), calibration analysis and decision curve analysis. The model was used for internal and external validation, and was presented as a nomogram. ResultsIn total, 6257 participants were recruited for this study. Age, malnutrition, chronic respiratory failure, body mass index, and activities of daily living scores were identified as the final predictors. The AUC of the model in the training and validation set was 0.87 (95 % confidence interval [CI], 0.85–0.89), 0.88 (95 % CI, 0.85–0.91), respectively. The model demonstrated acceptable calibration and clinical benefits. ConclusionsComorbidities in patients with cancer are closely related to the etiology of pressure injury, and can be used to predict the risk of pressure injury. Implications for practiceThis study provides a tool to predict the risk of pressure injury for cancer patients. This suggests that improving the respiratory function and nutritional status of cancer patients may reduce the risk of community-acquired pressure injury.

Jump Landing Kinematics: Establishing Normative Ranges For Male And Female Athletes.

Lower extremity joint (LE) kinematics during landing tasks are important predictors of injury risk and performance outcomes in athletes. To establish sex-related differences and normative ranges for LE kinematics during the jump-landing task in a large cohort of healthy military service academy cadets. Cross-Sectional Study. US Air Force, Naval, and Military Academies. 5308 cadets (2062 females [38.8%]). Sex-related differences in LE kinematics were analyzed using independent samples t-tests. Mean differences (MD) and effect sizes (d) were reported for interpretability. Normative ranges for hip and knee joint angles were established separately for males and females at initial contact (IC) and 50% of the stance phase. Compared to males, moderate effect sizes (d ≥ .5) were observed for knee external rotation (negative value) where females displayed greater motion at IC and at 50% stance (MD: - 3.9˚ and -5.0˚, respectively, p < .001). The following findings were of small effect size (.2 ≥ d > .5). Females exhibited less knee and hip flexion at IC (MD: -1.8˚ and -0.5˚, respectively, p < .001) and at 50% stance (MD: -4.1˚ and -4.6˚, respectfully, p < .001). This was accompanied by females having greater knee valgus (negative value) and hip adduction at IC (MD: -2.2˚ and 1.06˚, respectively, p < .001) and at 50% stance (MD: -3.2˚ and 1.8˚, respectfully, p < .001). This study establishes normative ranges for LE kinematics during the jump-landing task in a large cohort of healthy military service academy cadets entering their first year. Sex- related differences in LE kinematics were observed, highlighting the importance of considering sex as a factor in the evaluation of lower extremity movement quality and management of injury risk.

Predicting Injury Risk in Sports Students: A Gender-Based Review

Predicting Injury Risk in Sports Students: A Gender-Based Review

Predicting pressure injury risk in hospitalised patients using machine learning with electronic health records: a US multilevel cohort study

ObjectiveTo predict the risk of hospital-acquired pressure injury using machine learning compared with standard care.DesignWe obtained electronic health records (EHRs) to structure a multilevel cohort of hospitalised patients at risk...

A machine learning algorithm-based predictive model for pressure injury risk in emergency patients: A prospective cohort study

ObjectivesTo construct pressure injury risk prediction models for emergency patients based on different machine learning algorithms, to optimize the best model, and to provide a suitable assessment tool for preventing the occurrence of pressure injuries in emergency patients. MethodsA convenience sampling was used to select 312 patients admitted to the emergency department of a tertiary care hospital in Tianjin, China, from May 2022 to March 2023, and the patients were divided into a modeling group (n = 218) and a validation group (n = 94) in a 7:3 ratio. Based on the results of one-factor logistic regression analysis in the modeling group, three machine learning models, namely, logistic regression, decision tree, and neural network, were used to establish a prediction model for pressure injury in emergency patients and compare their prediction effects. The optimal model was selected for external validation of the model. ResultsThe incidence of pressure injuries in emergency patients was 8.97 %, 64.52 % of pressure injuries occurred in the sacrococcygeal region, and 64.52 % were staged as stage 1. Serum albumin level, incontinence, perception, and mobility were independent risk factors for pressure injuries in emergency patients (P < 0.05), and the area under the ROC curve of the three models was 0.944–0.959, sensitivity was 91.8–95.5 %, specificity was 72.2–90.9 %, and the Yoden index was 0.677–0.802; the decision tree was the best model that The area under the ROC curve for the validation group was 0.866 (95 % CI: 0.688–1.000), with a sensitivity of 89.8 %, a specificity of 83.3 %, and a Yoden index of 0.731. ConclusionsThe decision tree model has the best predictive efficacy and is suitable for individualized risk prediction of pressure injuries in emergency medicine specialties, which provides a reference for the prevention and early intervention of pressure injuries in emergency patients.

Non-contact lower limb injuries in Rugby Union: A two-year pattern recognition analysis of injury risk factors.

The cause of sport injuries are multifactorial and necessitate sophisticated statistical approaches for accurate identification of risk factors predisposing athletes to injury. Pattern recognition analyses have been adopted across sporting disciplines due to their ability to account for repeated measures and non-linear interactions of datasets, however there are limited examples of their use in injury risk prediction. This study incorporated two-years of rigorous monitoring of athletes with 1740 individual weekly data points across domains of training load, performance testing, musculoskeletal screening, and injury history parameters, to be one of the first to employ a pattern recognition approach to predict the risk factors of specific non-contact lower limb injuries in Rugby Union. Predictive models (injured vs. non-injured) were generated for non-contact lower limb, non-contact ankle, and severe non-contact injuries using Bayesian pattern recognition from a pool of 36 Senior Academy Rugby Union athletes. Predictors for non-contact lower limb injuries included dorsiflexion angle, adductor strength, and previous injury history (area under the receiver operating characteristic (ROC) = 0.70) Dorsiflexion angle parameters were also predictive of non-contact ankle injuries, along with slower sprint times, greater body mass, previous concussion, and previous ankle injury (ROC = 0.76). Predictors of severe non-contact lower limb injuries included greater differences in mean training load, slower sprint times, reduced hamstring and adductor strength, reduced dorsiflexion angle, greater perceived muscle soreness, and playing as a forward (ROC = 0.72). The identification of specific injury risk factors and useable thresholds for non-contact injury risk detection in sport holds great potential for coaches and medical staff to modify training prescriptions and inform injury prevention strategies, ultimately increasing player availability, a key indicator of team success.

A novel approach for sports injury risk prediction: based on time-series image encoding and deep learning.

The rapid development of big data technology and artificial intelligence has provided a new perspective on sports injury prevention. Although data-driven algorithms have achieved some valuable results in the field of sports injury risk assessment, the lack of sufficient generalization of models and the inability to automate feature extraction have made it challenging to deploy research results in the real world. Therefore, this study attempts to build an injury risk prediction model using a combination of time-series image encoding and deep learning algorithms to address this issue better. This study used the time-series image encoding approach for feature construction to represent relationships between values at different moments, including Gramian Angular Summation Field (GASF), Gramian Angular Difference Field (GADF), Markov Transition Field (MTF), and Recurrence Plot (RP). Deep Convolutional Auto-Encoder (DCAE) learned the image-encoded data for representation to obtain features with good discrimination, and the classifier was performed using Deep Neural Network (DNN). The results from five repeated experiments show that the GASF-DCAE-DNN model is overall better in the training (AUC: 0.985 ± 0.001, Gmean: 0.930 ± 0.007, Sensitivity: 0.997 ± 0.003, Specificity: 0.868 ± 0.013) and test sets (AUC: 0.891 ± 0.026, Gmean: 0.830 ± 0.027, Sensitivity: 0.816 ± 0.039, Specificity: 0.845 ± 0.022), with good discriminative power, robustness, and generalization ability. Compared with the best model reported in the literature, the AUC, Gmean, Sensitivity, and Specificity of the GASF-DCAE-DNN model were higher by 23.9%, 27.5%, 39.7%, and 16.2%, respectively, which confirmed the validity and practicability of the model in injury risk prediction. In addition, differences in injury risk patterns between the training and test sets were identified through shapley additivity interpretation. It was also found that the training volume was an essential factor that affected injury risk prediction. The model proposed in this study provides a powerful injury risk prediction tool for future sports injury prevention practice.

Development and validation of a machine learning-based fall-related injury risk prediction model using nationwide claims database in Korean community-dwelling older population

BackgroundFalls impact over 25% of older adults annually, making fall prevention a critical public health focus. We aimed to develop and validate a machine learning-based prediction model for serious fall-related injuries (FRIs) among community-dwelling older adults, incorporating various medication factors.MethodsUtilizing annual national patient sample data, we segmented outpatient older adults without FRIs in the preceding three months into development and validation cohorts based on data from 2018 and 2019, respectively. The outcome of interest was serious FRIs, which we defined operationally as incidents necessitating an emergency department visit or hospital admission, identified by the diagnostic codes of injuries that are likely associated with falls. We developed four machine-learning models (light gradient boosting machine, Catboost, eXtreme Gradient Boosting, and Random forest), along with a logistic regression model as a reference.ResultsIn both cohorts, FRIs leading to hospitalization/emergency department visits occurred in approximately 2% of patients. After selecting features from initial set of 187, we retained 26, with 15 of them being medication-related. Catboost emerged as the top model, with area under the receiver operating characteristic of 0.700, along with sensitivity and specificity rates around 65%. The high-risk group showed more than threefold greater risk of FRIs than the low-risk group, and model interpretations aligned with clinical intuition.ConclusionWe developed and validated an explainable machine-learning model for predicting serious FRIs in community-dwelling older adults. With prospective validation, this model could facilitate targeted fall prevention strategies in primary care or community-pharmacy settings.