Comparing the incomparable: Ranking motion recordings when accuracy data is not available

Many different types of biomedical data can be recorded, however not every method of interpreting this is equally applicable for comparison within or between test groups. When comparing related but dissimilar group members sometimes it is necessary to use alternative methods of analysis, potentially with data derived from the originally recorded data. This paper outlines one such method that was used to compare a biological arm to a prosthetic arm, without relying on values of accuracy. By deriving data that represents the reliability of both the prosthetic and biological arm, it was possible to determine a rating of how functionally “good” each arm is individually. These ratings were compared to allow a comparison of arm function, rather than the actual arms themselves. It was determined that of the arms tested, the prosthetic arm was less reliable than the biological arm. This method can be used to further compare these two arms to other arms. Future research could include the addition of multiple prosthetic arms, larger data sets and other methods of comparison.