Wearable Tibia Sensor Shows Solid but Imperfect Reliability for Tracking Running Impacts

Every stride a runner takes sends a shockwave up the leg, and for decades researchers have wanted a cheap, portable way to measure it. A new study published in Sports Engineering has put one of the most promising commercially available devices, the TgForce wearable accelerometer, through a rigorous reliability test, and the results offer both encouragement and a note of caution for the booming market of running wearables. The work, led by Renata Teles Pereira and colleagues at institutions in Brazil, is the first to examine the reliability and measurement error of this specific sensor for assessing tibial acceleration and cadence in recreational runners.

The motivation is straightforward. Running has surged in popularity worldwide thanks to its low cost and broad physical, mental, metabolic, and social benefits, but it comes with a stubbornly high injury toll. Prevalence figures for running-related injuries range from 19 percent to 79.3 percent, with novice runners, those with less than twelve months of practice, at greater risk than experienced ones. The knee, leg, and foot are the most commonly affected regions, and conditions such as medial tibial stress syndrome, Achilles tendinopathy, and plantar fasciitis dominate the clinical picture. Understanding the biomechanical loads that drive these injuries is therefore a central goal of sports medicine.

Two measurements have emerged as practical surrogates for the loading a runner’s body experiences with every footstrike: tibial acceleration, the peak shock recorded along the shin bone, and cadence, the number of steps taken per minute. Higher tibial acceleration values have been associated with greater average and instantaneous loading rates in injured runners, while lower cadence has been linked to higher braking impulse, greater loading rates, and increased work at the hip, knee, and ankle. Critically, both measures respond to intervention. Studies have shown that a 50 percent reduction in tibial acceleration reduced running pain in athletes with patellofemoral pain six months after gait retraining, and cadence increases of 7.5 to 10 percent decreased pain during running and daily activities in the same population.

The problem has always been measurement. The gold standard tools, laboratory-based motion capture systems and force plates, deliver precise kinetic and kinematic data but are expensive, complex to set up, and confined to specialized research facilities. Tibia-mounted wearable sensors have emerged as a portable alternative, allowing clinicians to monitor runners outside the laboratory and integrate biomechanical insights into rehabilitation and training programs. Wearable devices have shown promising accuracy for spatiotemporal parameters such as distance and speed, and the TgForce sensor was selected for the new study because, in a comparative test against a laboratory reference system, it was the only commercially available wearable among five devices tested to provide clinically meaningful estimates of instantaneous vertical loading rate.

To test whether the device could deliver consistent readings, the researchers conducted a cross-sectional study following the COSMIN guidelines for studies of health measurement instruments. Forty-three recreational runners, 33 women and 10 men with a mean age of 22.4 years, all habitual rearfoot strikers with no running-related injuries in the previous six months, ran on a motorized treadmill at a standardized speed of 10 kilometers per hour. A 10-gram triaxial accelerometer was attached to the anteromedial aspect of the distal third of the right tibia with an elastic band and Velcro strap, sampling data at 1000 hertz. The device, which requires no manual calibration because signal preprocessing is embedded in its proprietary algorithm, recorded peak vertical, sagittal, and resultant tibial acceleration in units of g, along with step rate.

The reliability design was thorough. Two trained examiners independently attached the sensor and took measurements on the same day to establish inter-rater reliability, probing the influence of examiner-dependent factors such as sensor positioning, alignment with the tibial longitudinal axis, and fixation. One examiner then repeated the full protocol seven days later to assess intra-rater reliability across sessions. Reliability was quantified using intraclass correlation coefficients, with values between 0.76 and 0.90 considered good and between 0.51 and 0.75 considered moderate, while measurement error was expressed as the standard error of measurement and the smallest detectable change, and Bland-Altman plots were used to hunt for systematic and proportional bias.

The headline results were largely positive. Inter-rater reliability was good across the board: 0.79 for vertical tibial acceleration, 0.85 for sagittal tibial acceleration, 0.81 for resultant tibial acceleration, and 0.88 for cadence. This means different evaluators, each attaching the sensor themselves, can obtain consistent measurements, a reassuring finding given that sensor placement is often cited as the weak link in wearable biomechanics. Intra-rater performance was more mixed. Vertical tibial acceleration again showed good reliability at 0.82, but sagittal tibial acceleration dropped to a moderate 0.65, resultant acceleration to 0.71, and cadence to 0.73. In other words, readings taken by the same examiner a week apart were somewhat less stable than same-day readings by different examiners.

The bias analysis revealed subtler problems that matter enormously for clinical use. Proportional bias, where the size of the measurement discrepancy grows with the magnitude of the value being measured, was identified for inter-rater measurements of vertical tibial acceleration and for intra-rater measurements of sagittal tibial acceleration. This is clinically significant because the runners with the highest impact magnitudes, precisely those of greatest interest for injury risk screening and gait retraining, are also those for whom the device shows the largest discrepancies. The researchers suggest that small residual variations in sensor positioning and orientation over the tibia, which are difficult to eliminate entirely, tend to have a greater influence on the recorded signal at higher acceleration magnitudes. Systematic bias was also found for intra-rater measurements of sagittal and resultant acceleration, while cadence showed no systematic or proportional bias in either analysis.

The measurement error figures carry their own lesson for anyone using such devices to track progress. The smallest detectable change ranged from 2.58 to 3.18 g for the acceleration measures, and a striking 23.86 steps per minute for cadence. Given that injured recreational rearfoot strike runners typically show tibial acceleration values of roughly 8 to 10 g, and that gait retraining programs commonly aim for reductions of 25 to 50 percent, larger acceleration reductions are likely to exceed measurement error and represent true biomechanical change. But cadence is a different story: healthy recreational runners at low speeds typically run at 160 to 170 steps per minute, and retraining programs target increases of 5 to 10 percent, corresponding to changes of roughly 8 to 17 steps per minute. Those modest changes may fall within the device’s measurement noise, so clinicians should avoid interpreting small cadence shifts as real improvements unless they exceed the reported threshold.

The authors also flag important limitations and practical recommendations. The study included only healthy, uninjured, rearfoot-striking young adults, did not control footwear, tested only a single device unit, and collected data on a treadmill under non-fatigued conditions. Previous research has shown that tibial shock tends to be higher during outdoor running than on a treadmill, with treadmill-based measures explaining only part of the variance seen in the real world, so the reliability estimates may not fully generalize to road and trail conditions. To minimize error, the team advises standardizing sensor placement between sessions and examiners as closely as possible, using the same device unit and attachment procedure when comparing across days, running familiarization trials, and averaging repeated measurements. Future studies, they conclude, should test the device outdoors, across different foot strike patterns, footwear types, fatigue states, and multiple sensor units. For now, the message is balanced: the sensor is a credible tool for monitoring running impacts, but its numbers deserve the same statistical scrutiny as any clinical measurement.

Subject of Research: Reliability and measurement error of a wearable accelerometer for tibial acceleration and cadence in recreational runners

Article Title: Reliability and measurement error of a wearable sensor for tibial acceleration and cadence in recreational runners

Article References: Reliability and measurement error of a wearable sensor for tibial acceleration and cadence in recreational runners. (n.d.). https://doi.org/10.1007/s12283-026-00571-6

Image Credits: AI Generated

DOI: 10.1007/s12283-026-00571-6

Keywords: wearable sensors, tibial acceleration, cadence, running biomechanics, gait retraining, running injuries, reliability, measurement error, accelerometer, sports engineering, recreational runners, injury prevention