Quantifying Anterior Cruciate Ligament Injury Resilience: A Screening and Composite Score Framework

Abstract

Background:

Over 1.4 million anterior cruciate ligament (ACL) injuries occur worldwide each year, but studies suggest that they may be preventable through targeted screening and training. Biomechanical factors from squatting, jumping, cutting, and running have been associated with ACL injury risk; however, current movement assessments are lengthy, limiting adoption and compliance. There is no consensus on which activities best assess modifiable biomechanical risk factors.

Purpose:

(1) To identify an optimal set of activities that explain an athlete's full-body, 3-dimensional biomechanical risk factors for ACL injury and (2) to build a framework to determine an individual's ACL injury resilience (AIR) score.

Study Design:

Descriptive laboratory study.

Methods:

The authors recruited adolescent female athletes who played soccer, basketball, or volleyball as their primary sport and had no history of injury. Data collection included 3-dimensional motion capture and ground-reaction forces during 5 common ACL injury risk screening activities and the computation of 35 biomechanical risk factors. Column subset selection identified the set of activities that best reconstructed the biomechanical factors from left-out activities. This activity set, with biomechanical thresholds from prospective studies of ACL injuries, was used to build the AIR score framework. The AIR score and the subscores for the trunk, hip, knee, and foot ranked participant biomechanical resilience. Statistical significance was evaluated at α = .050.

Results:

Twenty-seven females were included in this study (mean ± SD; age, 15.7 ± 1.2 years; body mass index, 22.19 ± 3.58 kg/m²). Two screening activities—run cuts and single-leg drop jumps—achieved the highest reconstruction performance, predicting the biomechanical factors from the left-out activities with 88.4% accuracy. The AIR score from this 2-activity set strongly agreed with an AIR score generated using all the information from 5 activities (r = 0.90; τ = 0.72; P < .001).

Conclusion:

The AIR screen and score provided a robust methodological and statistical framework that could be updated as future studies identify and characterize new biomechanical risk factors in larger cohorts.

Clinical Relevance:

The authors’ proposed screening provided an individualized evaluation of ACL injury resilience from just 2 activities, using an objective assessment of 3-dimensional movement. The AIR score framework identified athletes with vulnerable whole-body movement patterns, along with subscores that could be targeted with exercise interventions.

Keywords

anterior cruciate ligament injury screening assessment biomechanical risk factors injury prevention

Anterior cruciate ligament (ACL) injuries are a major concern, with an estimated 100,000 to 200,000 ruptures occurring annually in the United States and 1.4 million worldwide.^17,38 Female adolescent athletes experience a higher incidence of ACL injuries as compared with males, particularly in sports such as soccer, basketball, lacrosse, and volleyball.^4,8,31,36 ACL injuries often require surgery, increase the likelihood of posttraumatic knee osteoarthritis, and negatively affect an athlete's social well-being.^47,48 Importantly, many ACL injuries may be preventable since 72% to 88% are noncontact injuries linked to risky movement patterns.^2,14

Various biomechanical factors, such as limited knee flexion and high knee valgus moments, have been associated with increased risk of ACL injury.^13,20 Beyond the knee, limited trunk flexion, increased trunk lean,²¹ inadequate hip flexion,¹ restricted ankle dorsiflexion,¹⁵ larger ground-reaction forces, and increased joint moments^20,27,41 have also been associated with increased risk of injury. Several studies have identified thresholds, based on differences between uninjured and injured populations, linked to increased ACL injuries.^{3,12,20,21,26,27,41,50} Efforts to train and modify these biomechanical factors have yielded some success in changing movement patterns and reducing ACL injury incidence.^5,10,18,35 However, to optimize training efficacy, it is crucial to identify individuals who exhibit risk factors through screening.

Currently, no consensus exists on the most effective or comprehensive screening approach for assessing ACL injury–related biomechanics. Most screening studies focus on a single activity, neglecting the variety of ACL injury mechanisms, such as change of direction and single-leg landings.^16,25,33 The drop vertical jump and the associated Landing Error Scoring System (LESS) are frequently employed to screen athletes, scoring participants through visual assessment.³⁴ However, multisite evaluations suggest limited ecologic validity, with some studies showing no correlation to ACL injury risk.^23,28,42 Additionally, LESS neither captures kinetic risk factors, such as large ground-reaction forces and knee moments, nor quantifies kinematic risk factors, which require 3-dimensional motion analysis.⁴² Effective ACL injury screening would rapidly capture modifiable biomechanical risk factors across diverse context-relevant movements.

The purpose of this study was to develop and evaluate an assessment and scoring framework that quantifies ACL injury resilience (AIR) using objective, whole-body, 3-dimensional biomechanical risk factors across multiple activities. Specifically, we aimed (1) to identify an optimal set of activities that captures modifiable biomechanical risk factors and (2) to develop a score framework to rank participants based on known relationships between biomechanics and ACL injury risk.

Methods

We collected experimental data approved under institutional review board protocol 67713, employed musculoskeletal modeling to calculate biomechanical risk factors, and applied column subset selection (CSS) to identify the optimal set of screening activities and develop an assessment scoring framework (AIR score). We developed 3 scores using a combination of kinematic and kinetic factors from the optimal and full activity sets and compared them with Pearson correlation coefficients and Kendall's tau. Using representative participants with differing AIR scores, we visually compared musculoskeletal models during key activities with established movement profiles.

Experimental Data Collection

Female adolescent recreational athletes without a history of knee injury were enrolled in this cross-sectional investigation. Participants signed an informed consent or assent form depending on their age and participated in the study following institutional review board approval from April to July 2023. Athletes between the ages of 14 and 18 years were recruited if their primary sports were soccer, basketball, or volleyball and they had not competed professionally or received recruitment offers from collegiate varsity teams. Individuals with prior knee injuries requiring surgery were excluded.

Before data collection, participants warmed up for 5 minutes at a self-selected pace on a treadmill. Participants then performed 5 common ACL injury screening activities in a randomized order: single-leg squats, double- and single-leg drop vertical jumps, 90° drop cuts, and planned 90° run cuts (Supplemental Text, Supplemental Figure 1). Single-leg activities were performed on each leg. Athletes completed 2 sets of 5 repetitions of continuous squats and 3 repetitions of all other activities; they were allowed to practice each activity before data recording.

Kinematic data were recorded by a 16-camera motion capture system (Motion Analysis Corporation) at 200 Hz to track trajectories of 50 retroreflective markers. We adapted the marker set of Uhlrich et al,⁴⁵ replacing the sternoclavicular joint markers with a single sternal notch marker. In-ground force plates (Bertec Corporation) collected ground-reaction forces at 2000 Hz. We filtered experimental marker and ground-reaction force data using a fourth-order Butterworth low-pass filter. To retain 99.5% of the signal power in a fast Fourier transform power analysis, we selected cutoff frequencies of 4 Hz for squats, 30 Hz for drop jumps, 50 Hz for drop cuts, and 60 Hz for run cuts.

Biomechanical Modeling

To calculate kinematic and kinetic risk factors from the experimental data, we modified a generic 33-degrees-of-freedom musculoskeletal model^24,37 to include an additional knee adduction-abduction degree of freedom. We scaled this model to match each participant's dimensions during a static calibration pose and then used inverse kinematics in OpenSim³⁹ to compute joint angles, which we filtered at the same frequencies as the experimental data. We calculated external joint moments with inverse dynamics, using joint angles from inverse kinematics and filtered experimental ground-reaction force data.

For each activity, we analyzed data within a region of interest. For the 5 squats, we analyzed the middle 3 repetitions to avoid inconsistent motions at initiation and completion of the activity. Repetition start and end points were defined as the instant at which participants reached 95% of their maximum height. For drop jumps, drop cuts, and run cuts, we analyzed the time from foot strike (ground-reaction force >10 N) to toe-off (ground-reaction forces <10 N) on the force plate. We excluded trials if 2 feet contacted a force plate during single-leg trials or if foot contact was not entirely on 1 force plate.

From these regions of interest, we calculated 35 biomechanical factors associated with ACL injury risk across the activities (Supplemental Table 1), which included peak, range of motion, and initial contact values (peak during the first 50 milliseconds after foot strike). We normalized ground reaction forces to body weight and joint moments to percentage body weight × height. For single-leg activities, we treated trials from the dominant and nondominant limbs as independent, since there were no significant differences between the limbs (P > .05). We averaged factors across both limbs for double-leg activities.

Activity Selection

To identify an optimal set of ACL injury screening activities, we used CSS to measure how well biomechanical factors from activity sets linearly reconstructed biomechanical factors from left-out activities.⁴³ We identified an optimal set through a forward stepwise process, adding an activity and its associated biomechanical risk factors at each step (Supplemental Table 1).

First, CSS selected the single activity with the lowest average normalized root mean squared error (NRMSE) across all left-out factors. NRMSE was calculated by dividing the root mean squared error by the maximum observed value for each factor. Subsequently, CSS evaluated all 2-activity combinations, including the best activity, selecting the best-performing 2-activity set. This process continued with 3-activity combinations and so on, adding activities to the previous best set.

To address the uneven number of trials, with twice as many single- as double-leg trials, we employed a within-person resampling scheme to use all trials during model training.

We used leave-one-out cross-validation to select activities, training the linear model on 26 participants and testing on the one left out. The regression coefficients were used to reconstruct metrics from the left-out participant, which were compared with the trial averages of the left-out participant. This was repeated until all 27 participants were tested to obtain an average performance for each step.

Finally, we compared the best activity sets with Mann-Whitney U tests (α = .050). The set with the lowest NRMSE was defined as the optimal activity set. We repeated this process using only kinematic factors to identify the optimal kinematic activity set. This was computed in case kinetic measures are not available.

AIR Score Framework

We developed an AIR score framework to quantify biomechanical injury risk factors from screening activity sets identified with CSS. The AIR score uses biomechanical differences between injured and uninjured populations reported in previous studies to establish an AIR distribution (Figure 1).^{3,12,20,21,26,41,50} We demonstrated this framework with data from our uninjured cohort and published ACL injury thresholds. Table 1 references studies defining thresholds for each activity.^{3,12,20,21,26,41,50} Single-leg thresholds were from studies on relevant single-leg activities, while double-leg thresholds were obtained from studies with double-leg drop jumps.

Figure 1.

Anterior cruciate ligament injury resilience (AIR) score framework. Schematic representation of the AIR score calculation. Biomechanical risk factors from the study sample (green), ${\bar{x}}_{n}$ , were compared with literature-reported injured population values (red) to establish a sample-adjusted injury threshold (Equation 1). A resilience factor z score (Equation 2) was computed for each participant's biomechanical factor. We averaged the z scores across all factors for each activity to generate subscores for the trunk, hip, knee, and foot (Equation 3). The AIR score (Equation 4) is the mean of the subscores. ${\bar{x}}_{n}$ = sample mean for the nth resilience factor; $x_{n}$ = a participant's nth resilience factor; $σ_{n}$ = sample standard deviation of the nth factor; $t_{n}$ = literature threshold difference between injured and uninjured groups for the nth resilience factor.

Table 1

Study Sample Summary and Literature Benchmarks ^a

	Double-Leg		Single-Leg
Factors	Drop Jump	Literature Threshold	Squat	Drop Jump	Drop Cut	Run Cut	Literature Threshold
Kinematics
Trunk lean peak¹²	–	–	–	3.0 (2.9)^o	5.2 (3.9)^o	14.4 (4.4)^o	>1.3^o ^b
Trunk flexion IC²¹	24.0 (6.6)^o	<12.4^o	–	13.9 (7.3)^o	13.3 (7.0)^o	19.5 (7.7)^o	<12.4^o ^c
Hip flexion IC⁵⁰	–	–	–	40.4 (6.7)^o	37.9 (8.3)^o	40.7 (7.4)^o	<5^o ^d
Hip flexion ROM²⁶	74.2 (15.3)^o	<3.2^o	–	–	–	–	–
Knee flexion peak^3,20	101.0 (11.8)^o	<10.5^o	86.7 (10.7)^o	71.4 (8.1)^o	66.8 (8.4)^o	63.6 (5.3)^o	<21.7^o ^c
Knee valgus IC^20,41	0.2 (6.5)^o	>8.4^o	–	–0.9 (3.0)^o	0.6 (3.3)^o	4.3 (4.9)^o	1.4× ^{d , e}
Knee valgus peak²⁰	7.4 (4.1)^o	>7.6^o	–	–	–	–	–
Ankle flexion IC³	–	–	–	5.0 (3.4)^o	6.9 (3.6)^o	2.4 (5.2)^o	<12.2^o ^c
Kinetics
Hip flexion moment peak²⁰	13.3 (2.4) %BW × HT	>3.4 %BW × HT	–	–	–	–	–
Knee flexion moment peak²⁶	12.7 (1.4) %BW × HT	>1.2 %BW × HT	–	–	–	–	–
Knee valgus moment IC⁵⁰	–	–	–	5.4 (1.8) %BW × HT	8.2 (13.0) %BW × HT	6.0 (2.4) %BW × HT	>1.7 %BW × HT ^d
Knee valgus moment peak²⁰	1.7 (0.7) %BW × HT	>2.6 %BW × HT	1.86 (0.9) %BW × HT	–	–	–	–
vGRF peak²⁰	2.2 (0.4) BW	>0.3 BW	–	4.2 (0.5) BW	4.5 (0.5) BW	2.3 (0.3) BW	>0.6 BW

Data are presented as study sample mean (SD), ${\bar{x}}_{n}$ (σ_n), except for literature thresholds, t_n, which are reported as differences between the injured and uninjured groups from studies using 3-dimensional motion analysis (eg, motion capture or video-based methods). Double-leg activity thresholds are derived from studies analyzing double-leg drop jumps. Thresholds for single-leg activities are drawn from a combination of single-leg vertical hops, landing and cutting maneuvers, and side-cutting maneuvers. Dashes indicate not applicable. BW, body weight; HT, height; IC, initial contact; ROM, range of motion; vGRF, vertical ground-reaction force.

Single-leg triple vertical hop.

Landing and cutting maneuvers.

Side-cutting maneuver.

1.4-fold increase in risk of injury with each 1° increase in knee valgus angle at IC.⁴¹

Sample - adjusted injury threshold : {\bar{x}}_{n} \pm \frac{t_{n}}{2}

Equation 1

Resilience factor z score = \frac{x_{n} - ({\bar{x}}_{n} \pm \frac{t_{n}}{2})}{σ_{n}}

Equation 2

We adjusted our sample mean, ${\bar{x}}_{n}$ , based on ACL injury thresholds from the literature, $t_{n}$ , to generate a sample-adjusted injury threshold for each resilience factor (Equation 1). Then we computed a resilience factor z score (Equation 2) for each participant's factor to measure how one's performance deviated from the sample-adjusted ACL injury threshold, where $x_{n}$ is the participant's nth factor and $σ_{n}$ is the sample standard deviation of the nth factor. The peak knee valgus angle ACL injury threshold at initial contact during single-leg activities increased 1.4-fold per 1° increase of valgus.⁴¹ Thus, we used a standard z score multiplied by the threshold instead of a sample-adjusted ACL injury threshold.

A resilience factor z score of 0 means that a participant's biomechanical factors were equidistant from our sample mean and a literature-derived injured population mean (Figure 1). Positive z scores indicate greater AIR with biomechanical risk factors more similar to uninjured populations, whereas negative z scores suggest lower injury resilience with factors more similar to injured populations.

AIR subscores : \sum_{a = 1}^{b} \frac{1}{b} [\sum_{n = 1}^{m} \pm \frac{1}{m} (\frac{x_{n} - ({\bar{x}}_{n} \pm \frac{t_{n}}{2})}{σ_{n}})]

Equation 3

Using Equation 3, we calculated subscores for 4 body segments: trunk, hip, knee, and foot. For each, we first averaged the z scores across all m factors in each screening activity, a (Supplemental Table 1). Then, for each segment, d, we averaged z scores across all b activities that include factors in that segment to compute its subscore.

AIR score : \sum_{d = 1}^{4} \frac{1}{4} [\sum_{a = 1}^{b} \frac{1}{b} [\sum_{n = 1}^{m} \pm \frac{1}{m} (\frac{x_{n} - ({\bar{x}}_{n} \pm \frac{t_{n}}{2})}{σ_{n}})]]

Equation 4

Finally, we calculated the total AIR score (Equation 4) by averaging the subscores for the trunk, hip, knee, and foot. The AIR score reflects an athlete's overall performance, where a higher score indicates a more resilient form (less risk) and a lower score suggests a less resilient form (more risk).

Statistical Analysis and Composite Score Evaluation

We calculated 3 AIR scores: the AIR score using the kinematic and kinetic factors from the optimal activity set; the kinematic AIR (kAIR) score using only the kinematic factors from the optimal kinematic activity set; and the full AIR (fAIR) score using kinematic and kinetic factors from all activities. To evaluate the AIR and kAIR scores, we used Pearson correlation coefficients to compare their association with fAIR and the Kendall's tau test to assess the ordering of participant scores. For all tests, we used an alpha level of .050 to test for significance.

We compared visualizations of musculoskeletal models from biomechanical simulations of participants with low, moderate, and high AIR scores at key timepoints during activities in the optimal set. These visualizations and AIR scores were compared with literature-reported risky versus resilient movement profiles.

Results

Twenty-seven female recreational athletes participated in this study (mean ± SD; age, 15.7 ± 1.2 years; height, 1.66 ± 0.07 m; mass, 61.4 ± 10.7 kg; body mass index, 22.19 ± 3.58 kg/m²). Soccer was the primary sport for athletes (n = 18), followed by volleyball (n = 5) and basketball (n = 4). Twenty-four participants were right-leg dominant and 3 were left-leg dominant.

Optimal Screening Activity Set

CSS identified run cuts and single-leg drop jumps as the optimal activity set, with a predictive error of 11.6% (Figure 2A). Run cuts demonstrated the highest predictive value for a single activity, followed by single-leg drop jumps. Performance significantly improved with the addition of a second activity (P < .001), but including a third or fourth activity did not provide additional ACL injury–related biomechanical information and increased prediction noise. When only kinematic factors were used in this analysis, CSS selected all 5 activities as the optimal kinematic activity set.

Figure 2.

Anterior cruciate ligament injury resilience (AIR) screen and score. (A) Average normalized root mean squared error for activity sets selected by column subset selection. The optimal activity set (run cuts and single-leg drop jumps) is highlighted. Data are presented as mean (line), interquartile range (box), and range of lower and upper 25% of data (error bars). (B) Correlation of the full 5-activity AIR score, fAIR, with the AIR score derived from the optimal activity set. Participants 22 (red, low score), 6 (orange, moderate), and 15 (green, high) are shown as examples. (C) Correlation of the kAIR score, using only kinematic factors from the optimal kinematic activity set with all 5 activities, with the fAIR score using kinematic and kinetic data from all activities (r = 0.91).

AIR Score Correlations With fAIR

The AIR score, derived from kinematic and kinetic factors in the optimal activity set, strongly correlated with the fAIR score, the full 5-activity AIR score calculated from all factors and activities (r = 0.90; P < .001) (Figure 2B). All AIR and fAIR subscores were statistically correlated (trunk, r = 0.81; hip, r = 0.94; knee, r = 0.81; foot, r = 0.88; P < .001). Kendall tau indicated a strong significant correlation between the AIR and fAIR scores (τ = 0.72; P < .001), suggesting that each score ranked participants similarly.

kAIR Score Correlations With fAIR

The kAIR score, based only on kinematics, was also highly correlated with the fAIR score, based on kinematics and kinetics (r = 0.91; P < .001) (Figure 2C). All kAIR and fAIR subscores were statistically correlated (trunk, r = 1.00; hip, r = 0.96; knee, r = 0.91; foot, r = 0.74; P < .001). However, the foot subscore correlation was reduced, and participant ranking was moderately affected (τ = 0.54; P < .010) as compared with the AIR correlations.

AIR Score Distribution and Evaluation

Our healthy cohort exhibited positive AIR scores and was more aligned with uninjured population biomechanical risk factor values, as expected. The AIR scores in our cohort ranged from 0.1 to 1.4, with a mean score of 0.7 ± 0.4 (Figure 3). The AIR subscores highlight the performance of factors related to the trunk, hip, knee, and foot across all screening activities (Figure 3, B-E). Trunk and hip subscores during single-leg drop jumps and run cuts were lower than the knee and foot subscores: trunk, 0.5 ± 0.5 (range: –0.7 to 1.5); hip, 0.4 ± 0.8 (range: –1.1 to 1.8); knee, 0.7 ± 0.5 (range: –0.4 to 1.5); foot, 1.2 ± 0.7 (range: –0.2 to 2.8). The foot subscore had the highest mean, furthest from the sample-adjusted ACL injury threshold, but exhibited the most spread.

Figure 3.

Anterior cruciate ligament injury resilience (AIR) score participant rankings. Distribution of scores with participants ranked from low to high: (A) total AIR score and (B-E) subscores for the trunk, hip, knee, and foot. Three sample participants are color-coded to show the range of AIR scores: red (low), orange (moderate), and green (high).

We qualitatively assessed how the AIR score ranked participants by visualizing kinematics of 3 representative participants with low, moderate, and high ACL injury resilience during the run cut and single-leg drop jump (Figure 4). The participant with a low AIR score exhibited mechanics associated with increased risk and lower resilience to ACL injury. In contrast, the participant with a high AIR score demonstrated a more resilient and protective movement.

Figure 4.

Qualitative anterior cruciate ligament injury resilience (AIR) score evaluation. Visualizations of 3 participants’ musculoskeletal kinematics that represent varying AIR scores (low = red, moderate = orange, high = green). Comparisons for the optimal activity set were at (A) initial contact for the run cut and (B) maximum depth for the single-leg drop jump. Joint angles and center-of-mass locations that contribute to these subscores are marked on the figure.

Discussion

This study identified an optimal set of 2 activities—run cuts and single-leg drop jumps—to assess ACL injury resilience, which accurately reconstructed 3-dimensional biomechanical factors across 5 activities associated with injury risk. The AIR score framework used the optimal activity set to rank participants’ resilience to injury based on injury thresholds from prospective studies linking biomechanical factors to ACL injury.^{3,12,20,21,41,50} The AIR score from 2 activities performed comparably to a score using the full set of 5 activities.

Run cuts best reconstructed ACL injury–related factors from all other activities, and in fact, many noncontact injuries occur during cutting maneuvers.^9,11 Despite this, run cuts are underrepresented in controlled prospective studies comparing kinetic factors between injured and uninjured populations and are not included in other ACL injury risk screenings, such as LESS. Our work highlighted that run cuts contain meaningful information, supporting their inclusion in future assessments. Adding the single-leg drop jump as the second activity resulted in a small but statistically significant increase in accuracy. We also observed a stronger correlation between the fAIR and AIR (2-activity) scores than between the fAIR score and an AIR score with only run cuts (Supplemental Figure 2). These findings are consistent with prior research highlighting the importance of dynamic loading during single-leg landings.⁴⁴ Therefore, run cuts and single-leg drop jumps, identified as the optimal activity set, lay the foundation for the AIR screen and score.

Our qualitative assessment of the musculoskeletal model visualizations distinguished vulnerable and protective movement patterns, as described in previous studies, and provided insight into the validity of the scoring system.^19,22 The participant with a low AIR score demonstrated mechanics linked to ACL injury risk, such as excessive knee valgus, minimal knee flexion, and a posterior center of mass in the run cut—all of which increased knee moments—as well as a stiff landing, with minimal trunk, hip, and knee flexion in the single-leg drop jump.²² The participant with a moderate AIR score exhibited reduced hip flexion during run cuts and less knee flexion during single-leg drop jumps; however, she showed better knee alignment over the toes, resulting in a higher knee subscore, balancing risky and protective movement profiles. The participant with a high AIR score displayed protective biomechanics, characterized by greater sagittal-plane joint flexion, neutral frontal-plane lower limb alignment, and efficient force absorption, as demonstrated by her alignment of the trunk, knee, and foot during landing and cutting maneuvers.¹⁹ The AIR score's ability to capture these differences illustrates its effectiveness in discriminating among individuals based on their movement profiles. A strength of the AIR score framework is its ability to quantitatively integrate biomechanical information from multiple movement patterns, an advantage over current techniques that visually assess a single activity, such as the LESS and drop vertical jump assessments.

Moreover, advances in markerless motion capture using machine learning–based tools such as OpenCap and DeepLabCut may enable efficient, scalable AIR screening in schools and community settings, facilitating large-scale deployment of personalized AIR assessments and targeted intervention strategies.^29,45

The AIR assessment and composite score are not without limitations that affect the generalizability of results. While our initial results demonstrated the utility of the AIR score framework, we acknowledge that updating sample means, standard deviations, and injury threshold values with larger and more diverse longitudinal data sets that monitor injury outcomes associated with composite scores will improve specificity and sensitivity to injury risk classification across relevant populations. Additionally, lacking longitudinal ACL injury data for our cohort limited this initial version of the AIR score to rely on literature-reported injury thresholds. Prospective longitudinal data combined with surveillance of ACL injury outcomes could validate the AIR score, as well as update activity and biomechanical factor weightings for improved injury prediction. Additional data will also clarify the potential effects of limb dominance and athletic level.^30,46 Nonmodifiable factors also influence ACL injury risk, such as anatomic variations in intercondylar notch width, ACL material properties, and hormonal fluctuations.^6,7,32,40,49 While important, these factors were not included in this framework, as we focused on modifiable biomechanical factors that can improve with training. Finally, the AIR score, developed for uninjured individuals, is not intended for use with athletes after ACL reconstruction; alternative frameworks are needed to track recovery and evaluate return-to-sport readiness.

Conclusion

We introduced an AIR assessment and score framework, using run cuts and single-leg drop jumps, to evaluate an athlete's biomechanical performance. The AIR subscores assess ACL injury resilience at each joint and body segment. The AIR score framework can be expanded to incorporate more participant data and updated as future studies identify and characterize new biomechanical risk factors. This approach will help focus intervention efforts by identifying an individual's movement deficits. Combining the AIR screen with emerging video-based markerless motion capture technology enables large-scale longitudinal data collection, which is essential for improving identification of athletes with low ACL injury resilience, informing injury prevention training, and tracking performance changes over time.

Footnotes

Appendix

Acknowledgements

The authors thank the participants involved in this study as well as Parker Ruth, Jon Stingel, Dani Mendoza, and Josie Davidson, who were instrumental in their help with data collection.

Final revision submitted February 13, 2026; accepted February 21, 2026.

One or more of the authors has declared the following potential conflict of interest or source of funding: This work was supported by the Wu Tsai Human Performance Alliance and National Institutes of Health grant P41EB027060. AOSSM checks author disclosures against the Open Payments Database (OPD). AOSSM has not conducted an independent investigation on the OPD and disclaims any liability or responsibility relating thereto.

Ethical approval for this study was obtained from Stanford University (institutional review board protocol 67713).

ORCID iDs

Kirsten Seagers

Krithika Swaminathan

Julie Kolesar

Sam Hamner

Julie Muccini

Jennifer Hicks

Gregory Myer

Trevor Hastie

References

Blackburn

Padua

DA.

Sagittal-plane trunk position, landing forces, and quadriceps electromyographic activity. J Athl Train. 2009;44(2):174-179.

Boden

Dean

Feagin

Jr Garrett

Jr.

Mechanisms of anterior cruciate ligament injury. Orthopedics. 2000;23(6):573-578.

Boden

Torg

Knowles

Hewett

TE.

Video analysis of anterior cruciate ligament injury: abnormalities in hip and ankle kinematics. Am J Sports Med. 2009;37(2):252-259.

Bram

Magee

Mehta

Patel

Ganley

TJ.

Anterior cruciate ligament injury incidence in adolescent athletes: a systematic review and meta-analysis. Am J Sports Med. 2021;49(7):1962-1972.

Chappell

Limpisvasti

Effect of a neuromuscular training program on the kinetics and kinematics of jumping tasks. Am J Sports Med. 2008;36(6):1081-1086.

Chandrashekar

Mansouri

Slauterbeck

Hashemi

Sex-based differences in the tensile properties of the human anterior cruciate ligament. J Biomech. 2006;39(16):2943-2950.

Chandrashekar

Slauterbeck

Hashemi

Sex-based differences in the anthropometric characteristics of the anterior cruciate ligament and its relation to intercondylar notch geometry: a cadaveric study. Am J Sports Med. 2005;33(10):1492-1498.

Childers

Eng

Lack

, et al. Reported anterior cruciate ligament injury incidence in adolescent athletes is greatest in female soccer players and athletes participating in club sports: a systematic review and meta-analysis. Arthroscopy. 2025;41(3):774-784.e2.

Chinnasee

Weir

Sasimontonkul

Alderson

Donnelly

A biomechanical comparison of single-leg landing and unplanned sidestepping. Int J Sports Med. 2018;39(8):636-645.

10.

Cierson

Zhao

Belkhelladi

, et al. The effect of the FIFA-11+ ACL injury prevention program on drop vertical jump biomechanics in varsity athletes: a prospective observational cohort study. Orthop J Sports Med. 2025;13(4):23259671251333792.

11.

Cochrane

Lloyd

Buttfield

Seward

McGivern

Characteristics of anterior cruciate ligament injuries in Australian football. J Sci Med Sport. 2007;10(2):96-104.

12.

Collings

Diamond

Barrett

, et al. Strength and biomechanical risk factors for noncontact ACL injury in elite female footballers: a prospective study. Med Sci Sports Exerc. 2022;54(8):1242-1251.

13.

Dai

Mao

Garrett

Anterior cruciate ligament injuries in soccer: loading mechanisms, risk factors, and prevention programs. J Sport Health Sci. 2014;3(4):299-306.

14.

Della Villa

Buckthorpe

Grassi

, et al. Systematic video analysis of ACL injuries in professional male football (soccer): injury mechanisms, situational patterns and biomechanics study on 134 consecutive cases. Br J Sports Med. 2020;54(23):1423-1432.

15.

Fong

Blackburn

Norcross

McGrath

Padua

DA.

Ankle-dorsiflexion range of motion and landing biomechanics. J Athl Train. 2011;46(1):5-10. doi:10.4085/1062-6050-46.1.5

16.

Fox

Bonacci

McLean

Spittle

Saunders

A systematic evaluation of field-based screening methods for the assessment of anterior cruciate ligament (ACL) injury risk. Sports Med. 2016;46(5):715-735.

17.

Gordon

. Anterior cruciate ligament injuries. In: Garrick

, American Orthopaedic Society for Sports Medicine, eds. Orthopaedic Knowledge Update: Sports Medicine. 3rd ed. American Academy of Orthopaedic Surgeons; 2004:169-181.

18.

Hewett

Lindenfeld

Riccobene

Noyes

FR.

The effect of neuromuscular training on the incidence of knee injury in female athletes: a prospective study. Am J Sports Med. 1999;27(6):699-706.

19.

Hewett

Myer

Ford

KR.

Anterior cruciate ligament injuries in female athletes: part 1, mechanisms and risk factors. Am J Sports Med. 2006;34(2):299-311.

20.

Hewett

Myer

Ford

, et al. Biomechanical measures of neuromuscular control and valgus loading of the knee predict anterior cruciate ligament injury risk in female athletes: a prospective study. Am J Sports Med. 2005;33(4):492-501.

21.

Hewett

Torg

Boden

BP.

Video analysis of trunk and knee motion during non-contact anterior cruciate ligament injury in female athletes: lateral trunk and knee abduction motion are combined components of the injury mechanism. Br J Sports Med. 2009;43(6):417-422.

22.

Krosshaug

Nakamae

Boden

, et al. Mechanisms of anterior cruciate ligament injury in basketball: video analysis of 39 cases. Am J Sports Med. 2007;35(3):359-367.

23.

Krosshaug

Steffen

Kristianslund

, et al. The vertical drop jump is a poor screening test for ACL injuries in female elite soccer and handball players: a prospective cohort study of 710 athletes. Am J Sports Med. 2016;44(4):874-883.

24.

Lai

AKM

Arnold

Wakeling

. Why are antagonist muscles co-activated in my simulation? A musculoskeletal model for analysing human locomotor tasks. Ann Biomed Eng. 2017;45(12):2762-2774.

25.

Larwa

Stoy

Chafetz

Boniello

Franklin

Stiff landings, core stability, and dynamic knee valgus: a systematic review on documented anterior cruciate ligament ruptures in male and female athletes. Int J Environ Res Public Health. 2021;18(7):3826.

26.

Leppänen

Pasanen

Krosshaug

, et al. Sagittal plane hip, knee, and ankle biomechanics and the risk of anterior cruciate ligament injury: a prospective study. Orthop J Sports Med. 2017;5(12):2325967117745487.

27.

Leppänen

Pasanen

Kujala

, et al. Stiff landings are associated with increased ACL injury risk in young female basketball and floorball players. Am J Sports Med. 2017;45(2):386-393.

28.

Lisman

Wilder

Berenbach

Jiao

Hansberger

The relationship between Landing Error Scoring System performance and injury in female collegiate athletes. Int J Sports Phys Ther. 2021;16(6):1415-1425.

29.

Mathis

Mamidanna

Cury

, et al. DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat Neurosci. 2018;21:1281-1289.

30.

McPherson

Dowling

Tubbs

Paci

JM.

Sagittal plane kinematic differences between dominant and non-dominant legs in unilateral and bilateral jump landings. Phys Ther Sport. 2016;22:54-60.

31.

Montalvo

Schneider

Silva

, et al. “What’s my risk of sustaining an ACL injury while playing football (soccer)?” A systematic review with meta-analysis. Br J Sports Med. 2019;53(21):1333-1340.

32.

Moul

JL.

Differences in selected predictors of anterior cruciate ligament tears between male and female NCAA Division I collegiate basketball players. J Athl Train. 1998;33(2):118-121.

33.

Moustridi

Risvas

Moustakas

Predictive simulation of single-leg landing scenarios for ACL injury risk factors evaluation. PLoS One. 2023;18(3):e0282186.

34.

Padua

Marshall

Boling

, et al. The Landing Error Scoring System (LESS) is a valid and reliable clinical assessment tool of jump-landing biomechanics: the JUMP-ACL study. Am J Sports Med. 2009;37(10):1996-2002.

35.

Petushek

Sugimoto

Stoolmiller

Smith

Myer

GD.

Evidence-based best-practice guidelines for preventing anterior cruciate ligament injuries in young female athletes: a systematic review and meta-analysis. Am J Sports Med. 2019;47(7):1744-1753.

36.

Prodromos

Han

Rogowski

Joyce

Shi

A meta-analysis of the incidence of anterior cruciate ligament tears as a function of gender, sport, and a knee injury-reduction regimen. Arthroscopy. 2007;23(12):1320-1325.e6.

37.

Rajagopal

Dembia

DeMers

, et al. Full-body musculoskeletal model for muscle-driven simulation of human gait. IEEE Trans Biomed Eng. 2016;63(10):2068-2079.

38.

Samuelson

Balk

Sevetson

Fleming

BC.

Limited evidence suggests a protective association between oral contraceptive pill use and anterior cruciate ligament injuries in females: a systematic review. Sports Health. 2017;9(6):498-510.

39.

Seth

Hicks

Uchida

, et al. OpenSim: simulating musculoskeletal dynamics and neuromuscular control to study human and animal movement. PLoS Comput Biol. 2018;14(7):e1006223.

40.

Shultz

Kirk

Johnson

Sander

Perrin

DH.

Relationship between sex hormones and anterior knee laxity across the menstrual cycle. Med Sci Sports Exerc. 2004;36(7):1165-1174.

41.

Sigurðsson

Fl Pétursdóttir

Briem

The early peak knee abduction moment waveform is a novel risk factor predicting anterior cruciate ligament injury in young athletes: a prospective study. Knee Surg Sports Traumatol Arthrosc. 2025;33(5):1677-1685.

42.

Smith

Vacek

Johnson

, et al. Risk factors for anterior cruciate ligament injury: a review of the literature—part 2: hormonal, genetic, cognitive function, previous injury, and extrinsic risk factors. Sports Health. 2012;4(2):155-161.

43.

Sood

Hastie

A statistical view of column subset selection. J R Stat Soc Series B Stat Methodol. Published online May 16, 2025. doi:10.1093/jrsssb/qkaf023

44.

Taylor

Ford

Nguyen

Shultz

SJ.

Biomechanical comparison of single- and double-leg jump landings in the sagittal and frontal plane. Orthop J Sports Med. 2016;4(6):2325967116655158.

45.

Uhlrich

Falisse

Kidziński

, et al. OpenCap: human movement dynamics from smartphone videos. PLoS Comput Biol. 2023;19(10):e1011462.

46.

van der Harst

Gokeler

Hof

AL.

Leg kinematics and kinetics in landing from a single-leg hop for distance: a comparison between dominant and non-dominant leg. Clin Biomech (Bristol). 2007;22(6):674-680.

47.

van Niekerk

Matzkin

Christino

. Diversity in sports medicine and rehabilitation: psychological aspects of return to sport for the female athlete. Arthrosc Sports Med Rehabil. 2023;5(4):100738.

48.

Webster

Hewett

TE.

Anterior cruciate ligament injury and knee osteoarthritis: an umbrella systematic review and meta-analysis. Clin J Sport Med. 2022;32(2):145-152.

49.

Zazulak

Paterno

Myer

Romani

Hewett

TE.

The effects of the menstrual cycle on anterior knee laxity: a systematic review. Sports Med. 2006;36(10):847-862.

50.

Zebis

Aagaard

Andersen

, et al. First-time anterior cruciate ligament injury in adolescent female elite athletes: a prospective cohort study to identify modifiable risk factors. Knee Surg Sports Traumatol Arthrosc. 2022;30(4):1341-1351.