International Research Journal of Engineering and Technology (IRJET) Volume: 12 Issue: 11 | Nov 2025
www.irjet.net
e-ISSN: 2395-0056 p-ISSN: 2395-0072
OPTIMIZED MACHINE LEARNING MODELS FOR ACCURATE HEART DISEASE PREDICTION Mrs.R.Auxilia anitha mary1, Dr.T.Ramaprabha2 1Research Scholar,1Dept of Computer science, 1Nehru Arts and Science College Coimbatore, India. 2Associate Professor, 2Dept of Computer science,2Nehru Arts and Science College Coimbatore, India.
------------------------------------------------------------------------***------------------------------------------------------------------------Abstract— Enhanced predictive models are necessary for the early identification and intervention of heart disorders,
which continue to be a major global health concern. In order to anticipate cardiac sickness, this research suggests a deep learning-based method based on a sizable patient data set. By employing a deep neural network and a multi-layered neural network to identify intricate patterns in the data, the architecture improves forecast accuracy. The 14-attribute data set used in this investigation was taken from Kaggle. The long short-term memory (LSTM) and gated recurrent unit (GRU) deep learning models will be the main focus of this investigation. These models are put through a thorough review process with established performance measures, demonstrating their ability to discriminate between individuals with and without cardiac problems. The technique of deep learning outperforms conventional techniques and has encouraging prediction skills, according to the results. Including interpretable components increases the clinical value of the model and helps medical practitioners make more informed decisions.
Keywords- Machine learning, GRU, Heart diseases, LSTM. I. INTRODUCTION Preventive and accurate prediction models are critical for early management of cardiovascular diseases (CVDs), which are a leading cause of death and morbidity worldwide. Given the intricacy of cardiovascular health, using sophisticated deep-learning algorithms is a thorough method of examining a variety of patient data, such as clinical symptoms, medical history, and demographic data. The exponential growth of data over time limits the forecast accuracy of machine learning systems. To get more accurate data, the researcher plans to improve this conventional approach. This work evaluates and enhances the accuracy of cardiovascular disease prediction using a GRU and an LSTM. GRU and LSTMP were used to assess the effectiveness of the model that accurately predicts the existence or lack of heart disease [1]. The F1 score, recall, accuracy, and precision of the deep learning models LSTM and GRU are examined in this study. Consequently, it suggests the best model among them all. This study makes use of the cardiovascular data set from Kaggle. The data set is then ready for deep learning model training. Ultimately, the review process suggests the best model among all of them. To begin the heart disease prediction process, the variables are first taken out of the dataset and separated into training and testing data sets. The architecture, depicted in Figure 1, enumerates the six fundamental components that make up the proposed work. Each has a distinct function. Training data can be pre-processed using the GRU and LSTM algorithms, and then the data can be categorized for analysis of the patient's potential for heart disease. The outcome was generated with remarkable precision. By evaluating how well an LSTM and GRU recognize patterns in the dataset, this work seeks to enhance heart disease prediction models. Through the application of sophisticated deep learning algorithms, this work aims to improve early detection and preventative techniques for cardiovascular diseases by offering insights into the intricate relationships and interactions within the data. The results must also be analyzed, and new and enhanced models must be used in upcoming studies to increase accuracy. This study is divided into four sections: related work in Chapter 2, technique and methodology in Chapter 3, result and discussion in Chapter 4, and conclusion.
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 371