Classification of Respiration Episodes using Fuzzy Logic

Respiratory signals collected from young adults using Biopac´s abdominal strain gauge were properly filtered, amplified and digitized. An algorithm that combined Autoregressive (AR) and modified zero-crossing models was used to extract signal parameters such as the energy and frequency of the underlying signal. These parameters were used in a classification scheme based on fuzzy logic. Because of the variability of respiration signals, fuzzy logic provides a more natural classification as opposed to threshold based method [3]-[4]. Experimental results show that fuzzy logic presents a flexible and adaptable classificatory mechanism, which shows in percentage to which a segment of respiration signal belongs to one of the following categories: normal respiration, respiration with artifacts or apnea. It can be effectively used to reduce false alarms and improve classification of ambiguous cases.