Survey on data mining techniques in heart disease prediction
IEEEDesignCompetition
1. Monitoring of
Arduino-based
PPG & GSR Signals
through an
Android Device
IEEE EMBS ISC – DESIGN COMPETITION
Authors
Itaf Omar Joudeh (Computer Systems Engineering)
Mohamed Alhamwi (Communications Engineering)
Arslan Haq (Biomedical and Electrical Engineering)
2. Background
▪ Every year, 15 million suffer from heart attacks
▪ 6 million die
▪ heart attacks can lead to paralysis, and
loss of speech [1]
▪ If stress is left unmanaged, it can lead to:
▪ emotional problems, psychological problems,
physical problems, heart diseases, high blood
pressures, chest pains, or irregular heartbeats [2]
▪ Health care costs are soaring Figure 1: Heart Diseases’ Symptoms [3]
3. Objective and Impact
▪ Human Activity Monitoring (HAM) System
▪ facilitate medical attention, care, support
▪ vulnerable population (elderly, disabled, etc.)
▪ monitor heart rate variability
▪ factors (stress, age, gender, etc.)
▪ monitor irregular heartbeats (i.e. Arrhythmia)
▪ early detection of heart diseases’ symptoms
▪ monitor post heart attack or stroke recovery
Figure 2: Stress Factor [3]
Figure 3: Elderly People [3] Figure 4: Facilitating Care[3] Figure 5: Facilitating Support [3]
6. Results
▪ Extracted statistical parameters
from signals
▪ Three test experiments
1) Sitting
2) Standing
3) Running
▪ Looking for a correlation
between PPG and GSR
▪ linear fit
▪ coefficient of determination (𝑅2)
▪ As heart rate increases, the
mean GSR also increases. Figure 10: GSR Mean vs. Heart Rate
7. Closing Remarks
▪ Alternative solution to heart monitoring systems
▪ Reliable
▪ connection
▪ data collection
▪ Economical
▪ low energy
▪ cost-effective
▪ real-time
▪ portable
▪ The combination of PPG and GSR sensors is
relatively new
▪ Currently used in commercial wristwatches
▪ e.g. Empatica
▪ One’s wrist may not be the most appropriate location
to acquire a PPG signal
Figure 12: Arduino and Android [8]
Figure 11: Real-Time Monitoring [7]
Hello everyone, my name is Itaf Omar Joudeh and I am in Computer Systems Engineering. For the next 5 minutes or so, I will be presenting “Monitoring of Arduino-based PPG and GSR Signals through an Android Device”.
Every year, over 15 million people worldwide suffer from heart attacks, and nearly 6 million people die. A heart attack can also lead to disabilities such as paralysis, and loss of speech. Moreover, if stress is left unmanaged, it can also lead to heart diseases, and related problems. Recent studies have shown that changes in the levels of biological stress influence the heart rate. In fact, the pumping process of the heart and in turn the blood flow throughout the body may be affected by changes in biological stress.
The general motivation of this project was to develop a Human Activity Monitoring (HAM) device for a home-care system that:
facilitates medical attention, care, and support to the vulnerable population,
monitors heart rate variability in conjunction with stress levels, and
monitors irregular heartbeats (i.e. Arrhythmia).
As a result, the system would help in the early detection of heart diseases’ symptoms, and post heart attack or stroke recovery.
PPG is a commonly used technique to measure the changes in blood flow volume using light. PPG, being a biosignal, is commonly used to provide useful information related to human physiology. By analyzing a PPG waveform, parameters such as the heart rate, blood pressure, and blood oxygen saturation can be determined along with useful information. Acquisition of a PPG signal may be affected by artifacts such as power-line interference, sensor placement, and motion.
GSR is a technique used to measure the changes in the skin conductance, usually to indicate an estimate of the level of biological stress. Unlike a PPG signal, a GSR signal is aperiodic and a number of factors can significantly affect the nature of the response, such as age, temperature, humidity, and health.
Both PPG and GSR sensors can be used to indicate how stress affects the heart rate by a simultaneous analysis of their corresponding waveforms.
With regards to the proposed solution, a PPG sensor and a GSR sensor interface with an Arduino microcontroller to acquire the required signals. The PPG sensor is used for estimating the heart rate, while the GSR sensor is used for detecting the stress level from the patient’s fingertips. The Arduino microcontroller interacts with an Android application to monitor the two signals through a wireless serial Bluetooth Low Energy connection. A battery was used to power the system in order to make it portable.
Standard algorithms for deriving statistical parameters from PPG and GSR signals are available. These algorithms were adapted to the acquired signals, and converted into the Android application to help in finding a correlation between PPG and GSR signals.
The extracted statistical parameters were then used to find a correlation between PPG and GSR by means of linear fits and coefficients of determination. The three test experiments that were performed included sitting, standing, and running. At the end, it was found that as the estimated heart rate increased, the mean GSR value also increased. No other correlation (between the other parameters) was found.
This design provides an alternative solution to heart monitoring systems that is:
reliable in terms of connection and data collection,
economical,
real-time, and
portable
The combination of PPG and GSR sensors is relatively new. It is currently being used in commercial wristwatches such as Empatica. However, one’s wrist may not be the most appropriate location to acquire a PPG signal. According to research, placing a PPG sensor in areas like fingers, earlobes, toes, or under feet where there is a good uniform blood flow yields more reliable results.
With this I would like to conclude my presentation, and I am open for questions . . .