This three-part study investigated alternative pre-processing techniques to better understand the differences in patterns of ground reaction force (GRF) and load rate (LR) among runners with running-related injury (RRI). 534 runners were assessed on an instrumented treadmill with 3D kinematic data capture. Participants were classified as "injured" or "uninjured" and "rearfoot" (RF) or "non-rearfoot" (non-RF) strikers. The raw net GRF is normalized by time and then averaged across at least ten steps for the left and right foot; a double Gaussian characterizes the biphasic double-mass-spring model of running gait. Six parameters from the Gaussians were used to describe the relative differences and shape change based on injury condition. LRs were calculated using a central difference numerical derivative of the raw normalized net force data. 32% of runners reached peak negative LR (unloading) within the first 20% of stance. Injured RF strikers had 18% higher peak positive LR and 6% shorter time to peak than uninjured RF strikers (p < 0.05). Injured non-RF strikers showed peak negative LR 10% earlier in normalized stance, with a 10% shorter interval between positive and negative peaks (p < 0.05). The magnitude and timing of impact and active phases of GRF waveforms differed in runners with history of tibial stress fractures and current Achilles tendinopathy (p < 0.05). LR and LR timing are important in relation to specific RRI. These alternative pre-processing methods may help improve mechanistic understanding of GRF and LR and identify gait retraining foci for specific injury diagnoses in the future.
Running is a popular form of exercise and recreation worldwide, yet it carries a substantial risk of musculoskeletal injury, with incidence rates ranging from 3% to a staggering 85%. Despite advancements in running biomechanics research, the relationship between running-related injuries (RRIs) and kinetic variables such as peak ground reaction forces (GRFs) and load rates (LRs) is inconsistently reported. The literature suggests that higher GRF and LR may contribute to overuse injuries in bone and soft tissues. However, systematic reviews and meta-analyses provide mixed evidence, offering limited support for these models.
The inconsistency in findings can be attributed to several factors, including the controlled laboratory conditions for treadmill running, individual variation in foot strike patterns, footwear, runner characteristics, running speed, and injury timelines across studies. Recent research has emphasized the role of methodological factors, such as excluding non-vertical forces, pre-processing techniques like signal filtering, and limiting LR analysis to a narrow region of interest (ROI), in contributing to divergent conclusions drawn from kinetic data. Current analyses often focus on the vertical component and simplify GRF curves using low-pass filters with cutoff frequencies ranging from 20 to 50 Hz. This approach may reduce GRF metrics to discrete peaks and summarize LR values within a limited region of interest, which can be informative but necessarily eliminates much of the complexity of the GRF waveform.
Deeper insights could be provided by a more comprehensive analysis by including anterior-posterior and medial-lateral forces with the vertical component to form the net GRF. The net GRF was calculated as the vector resultant. Nonvertical forces are critical for neuromuscular control and the metabolic costs of running. Emerging evidence suggests that nonvertical force may be linked to fatigue, age, foot strike pattern, and injury. The apparent relevance of nonvertical force indicates that the current understanding of the relationship between GRF, LR, and various injuries may be incomplete.
The prevailing approach facilitates initial exploration, but studies show that there are likely underappreciated and meaningful GRF signals at frequencies above conventional cutoffs [30], and using low cutoff or unmatched lowpass filter methods can be problematic when interpreting physiological information across signals. For instance, Soft tissue vibrations in the lower extremity are significant in running and can be observed in high-speed video along with time series and frequency data from force plate measurements, but these signals are often lost or significantly modified in lowpass filtered data. Improving the accuracy of GRF and LR measurements requires processing methods that preserve detailed characteristics of the GRF waveform and may include methods such as conservative data averaging and Gaussian parameter determination. Expanding the conventional ROI to capture net GRF fluctuations and LR throughout the entire stance phase may provide a more complete, unique understanding of individual responses to ground impact and energy dissipation. From the clinical perspective, clarifying the GRF and LR patterns related to specific injury diagnoses and foot strike patterns would be a breakthrough. Sports medicine specialists would have more accurate data needed to improve therapeutic program design, gait retraining cues and developing running injury prevention programs.
This study used data to explore alternative pre-processing techniques by characterizing net GRF and LR patterns across the stance phase in endurance runners. The goal was to identify differences in GRF and LR responses during running based on injury status and type [bone, soft tissue]. We hypothesized that alternative preprocessing methods will reveal significant features in the LR associated with different types of injury. Additionally, we expect that modeling the system dynamics and using descriptive mathematics will describe the shape change of the GRF and show characteristic differences with injury status and type. To accomplish these aims, this study was comprised of three main parts: (1) part 1: fidelity analysis to compare filtering cutoff effects on GRF responses in a subsample of 15 runners, (2) part 2: a cross-sectional characterization of GRF and LR using the cutoff selected from part 1 across the entire cohort of 534 runners, and (3) part 3: comparative analysis using the methods developed in above among a subset of 29 injured and non-injured matched runners of different foot strike type.