Cardiovascular Diseases Detection Using Photo Plethysmography (PPG) Signal Data

,


Introduction
PPG has been demonstrated as a productive device for the quick diagnosis of disorders related to cardiac.By using PPG, the volume of blood fluctuations in the tissues has been measured.PPG serves as a promising technique for screening cardiovascular diseases.Circulation of blood from the heart to the toes is measured through PPG signals (Palanisamy, et.al, 2023).CVD causes leading death around the globe.The abnormalities of cardiac have caused 29.6% of death in worldwide during the year 2010 (Shabaan, M.,et.al, 2020) PPG is more accessible potentially which requires less training for measurements in providing accurate insights in real-time (Weng, W. H., et.al,2023).In the paper, we will be investigating CVD prediction using the data from Physio Net (PN).
The data on blood pressure provides clear signals for scheming cuff-less blood pressure to estimate algorithms.The matlab files (.mat) which contain the raw electrocardiogram (ECG), photoplethysmography (PPG), and arterial blood pressure (ABP) signals are stored as cell arrays%20 of matrices.
The purpose of this study is to enhance the classifier's performance in the detection of cardiovascular disease for PPG signals.The main purpose of this work is to compute the discrete cosine transform (DCT) for creating train and test sets by K-fold cross-validation.A classifier, linear regression, and neural network are used to pre-process PPG data.The effectiveness of the classifiers in recognizing the CVD from the PPG signal has been enhanced by the application of machine learning and deep learning approaches.

Methodology
The methodology used in this paper is explained as follows.

Data collection
The study's objective is to use a collection of subject features to distinguish between normal and diseased cardiovascular instances using the PPG signal.

Data Description 2.1 Age
Age is regarded as a major component that is closely connected to vascular stiffness and also has an impact on the PPG signal's form.

Gender
IT affects several bodily indications in people.According to Regitz-Zagrosek [4], men and women experience distinct CVD events including heart failure.

Height, Weight
This is used to the calculation of the Body Mass Index (BMI).They are the indicator of the overall health 2.4 Blood Pressure This is the key indicator for cardiovascular health.Any abnormalities in blood pressure, like hypertension, can easily indicate underlying cardiovascular issues.

Heart Rate
Changes in the heart rate can indicate abnormalities in heart function, they also indicate stress levels and physical activity.

Machine Learning and Deep Learning Classification Algorithm
K-fold cross-validation is a technique used in machine learning which helps in the Here we have used the Neural Network (NN) model which asses the PPG signal feature, by training this model on the dataset and predicting the likelihood of developing cardiovascular disease based on the PPG characteristics which is observed.The neural network is implemented using the framework known as TensorFlow.

Design Methodology
Detecting of cardiovascular disease step using PPG signal data involves a multi-process.

Data Collection
Data on patients have been collected, which includes details about gender, height, weight, blood pressure, and age.This helps in distinguishing between normal and abnormal cardiovascular cases.

Data Preprocessing
Preprocess includes reshaping the data, removing noises, and signal filtering which helps in analyzing the data without any disturbance.The overall process is depicted in figure 1.
Figure 1.The schematic diagram for the implementation

Feature Extraction
Feature extraction that is cross-correlation has been done as they help in aligning and synchronizing the PPG signals, overall, they help in extracting the relevant information about the PPG signals related to the prediction of cardiovascular.
Discrete Cosine Transform (DCT) is used in the signal's compression, noise removal, and dimensionality reduction which in turn helps to improve the quality of signal representation and extract the relevant information for accurate analysis and detection the cardiovascular diseases.

Model Selection
K-fold cross-validation is a technique used in machine learning that helps in the model selection, this selects the model that performs best on average across the K iterations, K-fold cross-validation helps in ensuring the model that is developed is more reliable and effective in detecting cardiovascular disease using PPG signal data.
Here we have used a neural network model which asses the PPG signal feature, by training this model on the dataset and predicting the likelihood of developing a cardiovascular disease based on the PPG characteristics which have been observed.

Model Training
The model is trained with the data rows, For the speed purpose, which is to train the model on a subset of the training data.

Prediction
The neural network implementation framework known as TensorFlow has been used.After using the models and then visualizing the predicted result will be 98% accurate

Results
PPG is a promising technique for early diagnosis of cardiac disorders, measuring blood volume fluctuations in tissues, and detecting cardiovascular diseases (CVDs).PPG signals are used to measure cardiac functions, such as blood flow, heart rate, and mean circulation time.PPG signals can be spectral analyzed to identify variations in heart rate, overall to detect the changes in blood volume levels, PPG serves as one of the efficient non-invasive, and simple techniques (Prabhakar, S. K., et.al , 2019).
Blood pressure (BP) is a bio-physiological signal that provides vital information about human health.High BP is a significant health risk factor, leading to various diseases such as heart disease, stroke, and kidney failure.Accurate prediction and measurement are crucial for diagnosis, prevention, and treatment (Stojanova, et.al. 2019).The high quality of the PPG signals can be measured by placing the fingers on the PPG device, PPG signal will require only less hardware which is more prominent compared to the electro diagram (ECG) (Ramachandran, et.al, 2020).
Cardiovascular disease has become the more common cause to death, and efforts in an early detection approach are a very effective way of reducing CVD (Ave, Arrozaq, et.al, 2015).Many methods have been adopted for the merging of model behavior, among them the linear regression model along with the neural network model has been used to obtain the analysis prediction Li, Gen, 2018).The plan and the execution of the PPG are very simple and have simple support, the depth analysis in detecting the cardiovascular disease (CVD) is done with the help of the PhysioNet database (Rajaguru, et.al., 2023).Photoplethysmography (PPG) assesses cardiovascular function using infrared light in peripheral areas.
PPG is a non-invasive and inexpensive method for monitoring patient physiological conditions, making it suitable for pulse oximetry, The main aim is to optimize the output of cardiovascular disease by using the PPG signal data by achieving the accuracy up to 98% (Rabhakar, et.al 2020).Neural network is an algorithm for detecting CVD using PPG signal data.It extracts relevant features like heart rate variability and pulse transit time as input variables and uses a binary label for detection.Neural networks, particularly deep learning models, have shown promise in healthcare particularly CVD detection.
Denoising PPG signals improves clinical prediction, especially for cardiovascular disease (CVD) identification.This method is independent of clinical analytics and demonstrates the potential of physiological signal pre-processing, specifically denoising, for improved performance (Ukil, Arijit, et al., 2016).

Figure 2 .
Figure 2. Visualization of Predicted values using linear regression