Heart disease is estimated to be the major cause of death among the middle-aged population worldwide. Researchers assess huge volumes of medical data using a variety of statistical, machine learning, and deep learning methods, supporting healthcare practitioners in predicting heart illness. This work aims to predict the likelihood of people developing heart disease using a wearable wristband that can record photoplethysmography (PPG) signals. Cardiovascular features extracted from the PPG signal are used to train the prediction algorithm. It enables the patient to self-monitor their health and take precautionary measures and treatment at the onset of symptoms of the disease. Random forest, convolutional neural network, long short-term memory networks are trained using publicly available databases comprising both affected and standard parameters and thereby used for comparison with the acquired sensor data for predictive analysis.
Top1 Introduction
Cardiovascular Disease (CVD) is a term that refers to a range of conditions that have an impact on the heart (Afilalo et al., 2009). They include lack of fitness, high cholesterol, hypertension, etc. An improper diet, harmful alcohol consumption, and excessive sugar levels are all factors that contribute to heart disease (Amin et al., 2013). Identifying and treating the people who are at risk of CVD is highly essential in order to reduce early deaths. Hence to prevent early death, an early and accurate medical diagnosis of heart disease must be made (https://appa.who.int/iris/handle/10665/43685). Lack of adequate medical datasets, lack of flexibility in selecting features, and lack of implementation of proper predictive algorithms are all obstacles that delay effective heart disease prediction (Sharma et al., 2020).
Males are more likely than females to get a CVD. It is twice as likely for males to develop heart disease during their lifetime as for females. The increased risk continued even when traditional heart disease risk factors such as lipid disorders, hypertension, insulin disorders, body mass index (BMI), and fitness activity were taken into account (Ebrahim et al., 1999). The factors that contribute to these differences in cardiac illness and mortality rates are numerous. Differential access and poor quality of health care, environmental or neighborhood impacts, persisting racial prejudice, health actions including nutrition, smoking, socioeconomic position, and genetic variation have all been proposed as factors contributing to congestive heart failure (Anderson et al., 1991).
ML and DL are examples of artificial intelligence approaches that can help with early screening and diagnosis of CVD at an early stage, in addition to prognosis evaluation and outcome prediction. With the proliferation of electronic health records (EHR), enormous amounts of quantitative, qualitative, and transactional data have been collected. With the use of clinically relevant information revealed in huge amounts of data, AI approaches can also assist clinicians in making the best clinical resolution, enabling prior diagnosis of subclinical organ dysfunction and therefore improving the quality and efficiency of cardiac healthcare (Faizal et al., 2021). A system based on such risk factors would not only assist medical experts and doctors but also alert patients to the possibility of CVD before they visit a medical center or undergo expensive medical examinations (Ebrahim et al., 1999). As a result, employing suitable classifier algorithms, this research proposes a technique for predicting heart disease using major risk variables. LSTM, CNN, and RF methods are among the major classification algorithms applied in this technique for analyzing whether patients have CVD or not.
In healthcare, machine learning and deep learning have shown effective assistance in assisting with decision-making and predicting from large datasets (Mohan et al., 2019). The machine learning disease prediction system uses information provided by users to identify diseases. Information entered into the web system predicts a patient's disease or the user's symptoms and gives results based on the information given. With its combination of ML and artificial intelligence, DL can be thought of as a means of simulating how humans acquire different types of knowledge. With the computer-aided diagnosis, this field relies on its own ability to learn and improve (Swathy & Saruladha, 2022).