An evaluation of the methods for deriving representative frequency response functions of the human whole-body system

The biodynamic response functions of the human whole-body system measured with subjects participating in an experiment are commonly arithmetically averaged and used to represent their mean response functions in many studies. The reported means were further averaged to form the reference means for standardization and various applications. The objectives of this study are to clarify whether this response-based averaging process could significantly misrepresent the characteristics of the original functions, and to explore appropriate methods for deriving representative functions. A set of reported mechanical-equivalent models for 12 subjects was used to derive the vertical and fore-and-aft cross-axis response functions expressed in apparent mass. The response-based average was directly compared with the response derived from a property-based derivation method. This study found that the response-based average could differ from the property-based mean response by more than 30%, especially in the fore-and-aft cross-axis response functions. This study also theoretically demonstrated that the discrepancies result from the non-linear relationship between the apparent mass and the properties of a dynamic system. Therefore, the discrepancies depend on the variability of the subjects' dynamic properties. Practically, the discrepancies in the vertical response could be reduced to an acceptable level (e.g., <10%) if a sufficient number of subjects with similar body weights are selected or grouped in the measurement. However, it is very difficult to reduce the discrepancies in the fore-and-aft cross-axis to such a level. While more demanding than the response-based method, the property-based method is theoretically more reliable for deriving the representative response functions for each axis.