Error rates are approximately three degrees when using a robotic knee [11], however these error rates then increase to 7.88 degrees with human use; due to sensor attachment issues and soft tissue movement [13]. In addition, these techniques require the patient to wear multiple sensors. There is a need to consider if clinically useful data can be obtained with fewer sensors, as for ambulatory monitoring applications, decreasing the number of sensors can greatly enhance the usability of the system [14]. The reported errors in estimating joint angular kinematics using inertial sensors are in a range which makes them accurate only for identifying large deviations to the gait pattern.

As some abnormal pathological gait patterns have only subtle deviations there is a need for an inertial sensor processing technique that can identify a pathological gait deviation in which no spatio-temporal differences exist, but only subtle kinematic differences exist.As such, the purpose of this work was to investigate innovative ways by which raw inertial sensor data could be processed to provide clinically useful information. Specifically we want to determine if inertial sensor extracted features can be used to identify abnormal gait patterns in a pathological population with subtle gait pattern differences.ACL-R gait was chosen as the gait pathology to investigate for three reasons. Firstly, aberrant gait patterns post-ACL-R have been suggested to be a potentially important risk factor for the development of knee OA [15�C17].

Secondly, ACL-R gait has not been previously investigated with inertial sensors and thirdly, the deviations between ACL-R and normal, healthy gait patterns are minimal [18�C20]. Therefore, ACL-R participants provide a novel test cohort to determine if inertial sensors can be used to detect gait changes that are not obvious to the eye.2.?Experimental SectionSeventeen lower limbs of fourteen female athletes constituted the ACL-R group. Of these athletes, three participants had previously ruptured both right and left ACL, thus both lower limbs were included for the analysis in these participants (Table 1). Of the seventeen involved lower limbs analyzed in this work, eight were reconstructed via a hamstring auto-graft surgical procedure, with the remaining being a bone-patellar tendon-bone auto-graft.

At the time of testing all athletes were fully engaged in field or court based sports (e.g., Gaelic Dacomitinib football, soccer, hockey, basketball) at club or county level and no athlete was undergoing any form of formal rehabilitation. Seventeen female athletes with no previous history of knee joint injury constituted the control group (Table 1). All athletes played field or court based sports (e.g., Gaelic football, soccer, hockey, basketball) at club or county level. Ethical approval for the study was approved by the Universities ethics committee.