Quality evaluation of ballistocardiogram from fiber optic sensors using fuzzy comprehensive evaluation method

Evaluating the quality of ballistocardiogram (BCG) is crucial for enhancing the accuracy of subsequent physiological parameter extraction. The micro-bend fiber optic BCG sensor and fiber Bragg grating BCG sensor are designed and manufactured, and experiments show that they can effectively acquire BCG. A BCG quality evaluation model based on fuzzy comprehensive evaluation is established, which uses four BCG evaluation indicators in time-domain and frequency-domain, including ratio of coefficient of variation, matching degree of J-wave detection, ratio of power spectral density, coefficient of template matching. By analyzing the quality of 800 manually labeled signal samples, the fuzzy membership functions for the four BCG evaluation indicators are determined. Considering the different noise features of various fiber optic sensors, different weight distributions are applied to BCG evaluation indicators for the micro-bend fiber optic BCG sensor and fiber Bragg grating BCG sensor. Applying the quality evaluation model to signals acquired at different times from two types of fiber optic sensors, and classifying them into excellent, average, and poor categories, the accuracy of BCG quality determination for the two types of sensors is 84.50% and 85.25%, respectively.