Respiratory Muscle Strength Assessment Algorithm Based On Cough Sounds In Patients With TSCI

With the rapid development of industry,the number of Traumatic Spinal Cord Injury(TSCI)patients is increasing,which brings great economic burden to the country and society.Respiratory failure is the main cause of early death.In clinical practice,pulmonary function instrument is often used to judge patients’ respiratory muscle strength,so as to diagnose patients’ respiratory system.However,using pulmonary function instrument to judge patients’ respiratory muscle strength is complicated and expensive,which brings great burden to patients’ families.Cough sound is used to judge respiratory muscle strength indirectly because its intensity is related to respiratory air flow.However,due to the influence of vocal tract modulation,even if the exhaled air flow is small,the sound signal with larger amplitude may be obtained after vocal tract modulation.In addition,TSCI patients often produce cries rather than real cough sounds due to discomfort,so the intensity of sound is also high when the patient does not exhale strongly.Therefore,it is easy to misjudge respiratory muscle strength directly based on the amplitude peak of cough sound in time domain or frequency domain.In view of the above problems,this paper studies the algorithm of TSCI patients’ respiratory muscle strength evaluation based on cough sound signal,and develops the corresponding Android application.The main contents include:1)Extraction of incentive signal: Based on the strong correlation between the measurements of the vocal tract modulation and cries at different times,inverse filters were designed using linear predictive coding to suppress the effects of cries and/or vocal tract modulation.The incentive signal generated by the exhaled air flow in the lungs is extracted from the cough sound signal,which can reflect the respiratory muscle strength.2)Establishment of respiratory muscle strength assessment model: Combined with the commonly used indicators for the assessment of respiratory muscle strength in TSCI patients and the generation mechanism of cough sound,the maximum envelope of incentive signal of cough sound and the energy in the burst stage were selected as features,and a model for the evaluation of respiratory muscle strength was established based on support vector machine.3)Design and implementation of respiratory muscle strength assessment software:Develop an application based on Android system that can collect cough sounds,display cough sound time domain waveform,play cough sounds,store cough sounds and help doctors assess respiratory muscle strength in patients with TSCI.In this paper,the effectiveness of the respiratory muscle strength assessment algorithm is tested by experiments.The test results of existing samples show that the accuracy and sensitivity of the algorithm can reach 94.36% and 98.09% respectively,and the functions of the Android application basically meet the needs of doctors,and achieve the expected effect.