Application of autocorrelation and cross-correlation analyses in human movement and rehabilitation research

Technical note.

To provide background theory and information and to describe relevant applications of autocorrelation and cross-correlation methodology as they apply to the field of motor control in human movement and rehabilitation research.

Commonly used methodologies for pattern and event recognition, determination of muscle activation timing for investigation of movement coordination, and motor control are generally difficult to implement, particularly with large datasets. A brief description of the underlying mathematical theory of correlation analyses is given, followed by 4 different examples of how this methodology is useful for research in the movement sciences.

Examples demonstrating the utility of correlation analyses are presented from several different studies conducted at the University of Waterloo.

Autocorrelation was used to demonstrate the presence of 60-Hz noise in an electromyography signal that was not visible in the raw data. A "top-down" paraspinal muscle activation pattern was demonstrated for healthy adults during gait, with the use of cross-correlation. Cross-correlation was also used to quantify coactivation of bilateral gluteus medius muscles during standing in individuals who developed low-back pain. Gender differences in gluteus medius control of mediolateral center of pressure were seen with the use of cross-correlation.

Autocorrelation and crosscorrelation have been shown to be an effective tool for several different applications in the movement sciences. Examples of the method's utility include noise detection within a signal, determination of relative muscle activation onsets for postural control, objective quantification of muscle coactivation, and relating muscle activations with mechanical events.