Machine Learning Algorithm for Respiration Analysis with Blind Source Seperation

Recent researches have shown that respiratory problems are increasing day by day. These respiratory disorders mainly affect aged people mostly during sleep. Various existing systems available are expensive and have many drawbacks such as they cannot be monitored in real time and they are non-continuous methods having contact with the user. They also depend on motion artifacts. To overcome these and to promote assistive living for elders a real time contactless respiration monitoring system is designed which uses a simple and easy way of monitoring using a pair of microphones. The microphone detects the exhaled airflow and analyzes the patient's condition. The method here used is Blind Source Separation which differentiates the respiratory signals using ICA algorithm. It also analyses the real time respiratory rate of the patient. Further disease classification is done using machine learning algorithm Naïve Bayes classifier. This system successfully tested in different environment. This system helps in easy diagnosis and early care of people and reduces life threatening situations.