Ventilation mode recognition using artificial neural networks

摘要: This study investigated the capabilities of artificial neural networks to identify spontaneous and pressure support ventilation modes from gas flow airway signals. After receiving written informed consent, waveforms were recorded 13 patients undergoing general anesthesia. During analysis, inspiratory phase each breath was extracted normalized in amplitude wavelength. Neural configured input flow, pressure, or both output ventilatory mode. network training accomplished with data 500 breaths obtained 7 patients. performance tested 433 remaining 6 Networks using recognized correctly 78% (337), 97% (423), 100% (433) test waveforms, respectively. Results indicate that can be used effectively for breathing pattern recognition encourage application other types respiratory problems.