Network with Weighted Fuzzy Membership Functions (NEWFM) [5][6][9]. Three-axis acceleration sensors constructed by vertically connecting two 2-axis acceleration

This paper proposes a method to detect falls using a Neural Network with Weighted Fuzzy Membership Functions(NEWFM) and wavelet-based feature extraction. Data extracted from study subjects were applied to the NEWFM after going through preprocessing and wavelet processes. From the created wavelet coefficients, 40 initial features were obtained using statistical methods, including frequency distributions and the amounts of variability in frequency distributions. In order to assess the fall detection performance of the NEWFM, the data were divided into training sets and test sets in ratio of 5:5 to conduct experiments with the respective ratios. Based on the results, the sensitivity, specificity and accuracy of the NEWFM were shown to be 98.02%, 98.99% and 98.5% respectively when the ratio of the training set and the test set was 5:5. Keywords—Fall Detection, Wavelet Transform, NEWFM, Signal Processing.