Mechanical loading prediction through accelerometry data during walking and running

Abstract

Currently, there is no way to assess mechanical loading variables such as peak ground reaction forces (pGRF) and peak loading rate (pLR) in clinical settings. The purpose of this study was to develop accelerometry-based equations to predict both pGRF and pLR during walking and running. One hundred and thirty one subjects (79 females; 76.9 ± 19.6 kg) walked and ran at different speeds (2-14 km•h −1 ) on a force plate-instrumented treadmill while wearing accelerometers at their ankle, lower back and hip. Regression equations were developed to predict pGRF and pLR from accelerometry data. Leave-one-out cross-validation was used to calculate prediction accuracy and Bland-Altman plots. Our pGRF prediction equation was compared with a reference equation previously published. Body mass and peak acceleration were included for pGRF prediction and body mass and peak acceleration rate for pLR prediction. All pGRF equation coefficients of determination were above 0.96, and a good agreement between actual and predicted pGRF was observed, with a mean absolute percent error (MAPE) below 7.3%. Accuracy indices from our equations were better than previously developed equations. All pLR prediction equations presented a lower accuracy compared to those developed to predict pGRF. Walking and running pGRF can be predicted with high accuracy by accelerometry-based equations, representing an easy way to determine mechanical loading in free-living conditions. The pLR prediction equations yielded a somewhat lower prediction accuracy compared with the pGRF equations.

Using Raw Accelerometer Data to Predict High-Impact Mechanical Loading

“…Nevertheless, these MAPE values are similar to the results usually found in accelerometer-based energy expenditure prediction models [ 37 ], which have widespread use in research. As for the pLR prediction equations, they presented a lower accuracy compared to the pGRF equations, corroborating previous findings [ 19 , 21 , 22 ]. This is possibly due to the fact that the loading rate is typically calculated through a derivative of the force signal from the beginning of the foot’s contact with the floor to the point where the pGRF is achieved [ 19 , 21 , 22 , 38 ].…”

“…When comparing the accuracy indices of the pGRF prediction equations developed in the present study with other equations from past investigations, developed for walking and running, it can be seen that our models have a lower accuracy [ 17 , 18 , 20 , 21 , 22 ]. This can be explained, at least partially, by the high heterogeneity in the jumping movement patterns, as several combinations of jump types and heights were tested in our study.…”

“…In the last few years, objective methods based on wearable devices have been proposed as a solution to overcome these limitations, since acceleration and some mechanical loading variables resulting from body movement have shown to be correlated [ 15 , 16 ]. In fact, several studies using different prediction model approaches (e.g., linear regression, linear mixed models, machine learning) have already shown that, through the use of acceleration data, it is possible to estimate some mechanical loading components that are typically assessed only through force plates, namely, peak ground reaction force (pGRF) [ 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 ] and peak loading rate (pLR) [ 19 , 21 , 22 ]. Although accelerometry-based equations have shown high prediction accuracy [ 21 , 22 ], this versatile approach to mechanical loading assessment is not widely used yet, either in clinical practice or research.…”

“…In fact, several studies using different prediction model approaches (e.g., linear regression, linear mixed models, machine learning) have already shown that, through the use of acceleration data, it is possible to estimate some mechanical loading components that are typically assessed only through force plates, namely, peak ground reaction force (pGRF) [ 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 ] and peak loading rate (pLR) [ 19 , 21 , 22 ]. Although accelerometry-based equations have shown high prediction accuracy [ 21 , 22 ], this versatile approach to mechanical loading assessment is not widely used yet, either in clinical practice or research. One of the issues that may contribute to this fact is the lack of validation studies covering a broader spectrum of activities, especially those known to have a greater osteogenic effect, such as high-impact activities.…”

“…Until now, studies that have calibrated accelerometers to predict mechanical loading variables have usually employed only simple activities, such as walking [ 19 , 20 , 21 ] or walking and running [ 17 , 18 , 22 , 24 ], while activities eliciting higher impact loading have not yet been tested. The fact that these accelerometry-based prediction equations are only valid for a limited set of activities hinders their ability to accurately quantify the amount of loading to which a person is subjected daily, as the physical activities performed daily are often composed of other types of activities than just walking and running.…”