Abstract
In this work, we present a vital sign monitoring system that can measure a person’s respiration rate to detect different stages of apnea. Respiration rate is one of the most important vital signs as a person’s respiration rate can be used to detect multiple diseases such as asthma, anxiety, pneumonia, cardiovascular and lung disease. Respiration rate can also be used to detect if a person is using narcotics or any drugs but for this project, we are reducing the scope to detect different stages of sleep apnea. A respiration rate is the number of breaths taken per minute and an average adult person’s normal respiration should be in between 12 to 20 breaths per minute while resting and if a person’s respiration rate is under 12 or over 25 breaths per minute while resting is generally considered as abnormal respiration rate. In this project, the abnormal respiration rate is sleep apnea and apnea is detected when the respiration rate drops to zero for ten seconds. Apnea is considered as mild apnea if the person’s respiration rate drops to zero for 5 to 14 episodes of 10 seconds in an hour, moderate apnea is considered if a person’s respiration rate drops to zero for 15 to 29 episodes of 10 seconds within an hour and severe apnea is considered if a person’s respiration rate drops to zero for 30 or more episodes of 10 seconds within an hour. To measure accurate respiration rate and detect different stages of apnea using conventional methods such as stethoscope checkup or capnography involves many challenges as it requires the physical presence of the patient in the examination room which is difficult for the elderly and disabled people. Some patients may not be comfortable with the physical contact making the respiration rate monitoring process harder. Furthermore, a patient may intentionally reduce the breathing rate while examination because of nervousness or any other reason which may lead to capturing the inaccurate results. To overcome such challenges a non-invasive and passive monitoring system is developed in this project to capturing respiration rates using remote photoplethysmogram and commodity WIFI based on the chest movements of the patient. The collected dataset contains both the normal breaths per minute as well as abnormal breaths per minute. Artificial intelligent algorithms such as SVM, KNN, and XGBoost are used to predict different stages of apnea.