Rosebrugh Bldg, Toronto, ON M5S 3G9
Room: RS 211
Falls due to slips are a major cause of injury in young and older adults, especially in the winter. Slip resistant footwear can prevent the occurrence of slips by providing traction. However, current standard mechanical tests perform poorly in measuring footwear slip resistance for winter conditions. This is due to the lack of standardized test surfaces for winter conditions and the negligence of how people walk. To address these limitations, a human-centered, objective test known as the Maximum Achievable Angle (MAA) test, was developed. Although it has higher accuracy, it can be further improved if additional information, such as the types of slips occurred and the associated slip distances, becomes available. In addition, it relies on visual detection of slips by a human observer, which is error-prone.
This study aims to develop a software algorithm that uses kinematic and foot pressure signals to detect, classify, and measure heel slip and toe slip, thereby improving the MAA test. It has three phases. In Phase 1, walking data from participants in a simulated winter environment was collected. Features identifying types of slips were extracted and used to train a machine learning algorithm, which are developed in Phase 2. In Phase 3, the algorithm will be validated with additional participants. The results from the algorithm will be compared with the actual results obtained by qualitative visual inspection of biomechanical data. Currently, we are labelling the walking data for machine learning. We expect that the algorithm can accurately detect, classify, and measure slips better than humans. If so, it will eliminate the dependence of the MAA test on human observer and improve its accuracy and resolution. Such improvements can facilitate standardization of the MAA test and lead to development of safer winter footwear.