<p>Left ventricular hypertrophy (LVH) is an analytical aspect of cardiovascular actions that might be perceived by echocardiography (ECHO) in an early phase. Early detection of LVH can be helpful for enhancing the outcomes of patients, which enables timely involvement and management of the state. In this framework, electrocardiography (ECG) acts as a perfect modality for tedious screening and observing of LVH. At present, few research works have been donated by machine learning (ML) for the ECG features to perceive the occurrence of LVH. Between these models, deep learning (DL) systems are greater than traditional ML models owing to their capability of automated feature extraction. This paper develops an Efficient Lightweight Model for Left Ventricular Hypertrophy Detection with Ensemble Learning and Seagull Optimization Algorithm (LMLVHD-ELSOA) in ECG Signals. The main objective of the LMLVHD-ELSOA algorithm is to evaluate the effectiveness of hybrid deep-learning models and optimization algorithms for ECG-based cardiovascular disease diagnosis. Initially, the presented LMLVHD-ELSOA model employs denoising and noise localization is performed using a lightweight improved U-Net model with an encoder-decoder structure. For the LVH diagnosis process, the proposed LMLVHD-ELSOA model deploys an ensemble of three models namely the bi-directional long short-term memory (Bi-LSTM) model, temporal convolutional network (TCN) approach, and wavelet neural network (WNN) classifier. Finally, the hyper parameter range method is implemented by an improved seagull optimization algorithm (ISOA) for enhancing the classification results of ensemble techniques. To validate the good classification result of the LMLVHD-ELSOA system, a wide sort of simulations occurs on benchmark MRI datasets. The wide range of comparative outcomes certified the furtherance of the LMLVHD-ELSOA system over the recent techniques.</p>