ABSTRACT
In today's world, everyone's health is a major concern and a top priority. Humans are afflicted with a plethora of diseases because of their unhealthy habits. People are primarily affected by heart attacks and low oxygen levels because of poor medical care and late diagnosis. As a result, this work aims to combat such untimely deaths using smart health monitoring, which employs machine learning and IoT. The proposed system includes ThingSpeak cloud to communicate with the doctor in case of any emergency. This system consists of body temperature sensor, pulse oximeter sensor (for collecting heartbeat rate and oxygen level) and blood pressure sensing module for tracking patient's health. These sensors are interfaced with the Raspberry pi and Arduino Uno microcontroller. The obtained result from patients is continuously monitored and it is updated in LCD and doctor's webpage using Internet of Things. Following these steps, a trained Machine Learning model is used to determine the type of disease being experienced by the patient. This system predicts Normal and two major disease namely Hypertension and Lung disease. By incorporating all these features, we can ensure that people who suffer from heart attacks and lung disease will not die suddenly. The accuracy of this proposed method is 86% approximately in a real time scenario. Furthermore, because raw medical data can be analyzed in a short period of time, the work will aid clinicians in remote monitoring during epidemic situations such as covid. © 2021 IEEE.
ABSTRACT
The objective of the proposed work deals with assisting the doctors by providing the required pre-diagnosis data of COVID-19 patients using radiology images of the targeted patients. A machine learning approach is utilized to evaluate the radiology images of COVID-19 patients which performs preprocessing, segmentation, feature extraction and classification. The proposed work also deals with predetermined evaluation results of COVID-19 patients which give the stages of the COVID-19 patients stating from STAGE 1 through STAGE 4. It will help the physicians for the easy diagnosis of the COVID-19 in patients. A conventional machine learning approach based classification is performed in first pass which discriminates the patients as normal patients or COVID-19 positive patients. Initially, the pre-processing of the input radiology images is carried out to enhance the quality of the images. Second order statistical textural features are extracted using Gabor filter bank Finally, the extracted features are used to classify the COVID-19 positive and negative patients using a simple decision tree classifier. During second pass, the affected portion of the lung is segmented, and amount of infection is estimated through the evaluation of length and width of the affected lung portions. Now, the COVID-19 positive cases will be given higher priority to undergo second level of diagnosis and treatment processes by the physicians whereas the COVID-19 negative cases will undergo for continuous observation. Thus, the proposed diagnosis system will help the physician to speed up their service towards COVID-19 positive patients. © 2021 IEEE.