Blood Viscosity based Heart Disease Risk
Prediction Model in EdgeFog
Abstract : In this paper, heart disease prediction modeled using partially observable Markova decision process (POMDP) is proposed. In emergency, the patient is alerted through the doctor by fog computing. Ambulance sent to the location of patient at critical situations. The doctor gets the data through fog computing iFogSim. Fog computing in healthcare is a new area, which gains more attraction in research community. Many researches focus on cardiovascular disease i.e. heart disease. The important risk factor for cardiovascular disease is increase in blood viscosity. The highly viscous nature of blood does not allow the blood to flow creating a resistance in the blood flow. Heart disease risk factors are high blood pressure, obesity, diabetes, increased blood viscosity, etc. With the help of POMDP’s states, observations, beliefs, probability transitions the patient health is noted. The POMDP model for heart disease prediction computes the policy approximation using states and timeslots. Rewards are tabulated using policy approximations over different iterations.
the number of works in the healthcare domain has dramatically increased to explore several undiscovered aspects or overcome existing drawbacks in health monitoring systems. With the purpose of novel health monitoring systems provision, some of works try to expand current systems with more services and functions while others attempt to propose new platforms or methods. However, as described in the aforementioned examples, a large number of systems focus on ZigBee whereas it is a challenge to assure the quality of service in ZigBee when monitoring streaming bio-signals such as ECG, EMG and Electroencephalography (EEG) as the maximum data rate of ZigBee is 250 kbps. Inversely, Bluetooth technology can overcome problems of low data rate in ZigBee and other short range communication protocols.
Existing healthcare systems that are deployed on IoT driven Fog or cloud computing frameworks connect pre-configured devices
for patient data processing such that the results are delivered to users within the deadline time. Many prior works have tried to use IoT to predict health problems related to heart but are unable to ascertain with the accuracies required by the stringent regulations of medical standardization agencies. In recent past, as deep learning has gained popularity more recent technologies can even surpass doctors in heart disease detection accuracy .
The proposed model uses classification algorithms for the diagnosis and prediction. The ensemble method of tree-based classification-Random Forest gives an accuracy of 93%. Sugondo Hadiyoso et al. proposed a mini wearable ECG device and real-time arrhythmia detection based on android mobile application. ECG signals can be captured by using the ECG's analog front end and sent to Android mobile through a Bluetooth module device. On Android application, data analysis can be done with the help of Pan Tompkins algorithms to detect complex QRS ECG signal and heartbeats.
Heart disease is caused because of an electrical breakdown in the cardiac signal of the heart. In this particular disease person loses consciousness and has no pulse which occurs death within a minute, leads to sudden cardiac arrest. The P, QRS complex and T wave of ECG signal triggered and generates an improper electrical signal that provides clinical information to diagnose. K. Amtul
Salam et al. introduced new technologies and algorithms to detect and analyze the ECG signal value.
|