The following key contributions are presented:

1. Proposes a low-power, low-cost wearable seizure detec tion system using integrated PIR+ACC sensors suitable for edge computing. 

2. Compares a probabilistic model (HMM) with a hybrid deep learning model (OptiNet-SVM combining CNN + SVM)for classifying Epileptic Seizure (MES), Normal Movement (MNMV), and No Movement (MNM). 

3. Utilizes advanced feature extraction techniques, includ ing AP+ML, DWT+DFT, and Non-Negative Spectrum Factorization (NNSF), to improve classification robust ness and detection latency. 

4. Achieves high accuracy (99.02%) and low latency (0.38 s) with ROC-AUC analysis confirming strong seizure separability, highlighting suitability for real-time clinical deployment. 

The study is under developing and validating an AI-based two-stage CMR interpretation system that accurately screens and diagnoses of cardiovascular diseases, outperforming cardiologists in some cases and offering a scalable solution for efficient CVD assessment.