Heart disease is the largest cause of patient death and has become a major public health problem in recent years. The main risk factors for heart disease are excessive alcohol use, smoking, and a sedentary lifestyle. In medical terms, heart illness is acknowledged as being difficult to diagnose, identify, and predict. Hospitals and other clinics are treating cardiac disorders with expensive procedures and treatments. As the population increases, diagnosis gets more challenging. Therefore, early detection of cardiac disease will help individuals worldwide by enabling them to receive appropriate therapy before the condition worsens. This research analyzes various feature selection and classification algorithms that have been developed for efficient cardiac disease prediction. The cardiac dataset is considered as an input data in this paper for analysis. It is initially pre-processed with different missing value imputation and normalization algorithms to replace the missing variables and standardize the raw data. These pre-processed data are then given to the feature selection process for selecting the optimal features. Finally, the selected features are given to classification approaches for effective prediction of heart disease. Among various feature selection algorithms Cheetah optimization (COA) produces better results than other optimization such as 93% F1_Score and 12% FOR. Then for accurate prediction of heart disease, LSTM algorithms achieve better performance as 95.05% accuracy and 93.59% precision. The experimental results show that COA-based feature selection and LSTM-based classification produce better results than other algorithms.