Alzheimer's disease (AD) is a progressive neurological syndrome which requires early and precise diagnosis in order to be effectively treated. Using multidimensional non-imaging clinical data that includes demographic, cognitive, lifestyle, and medical history aspects, a combined Deep learning and Machine learning framework is represented in this study to predict Alzheimer's disease. Correlation matrix was used to find inter-feature relationships and eliminate redundancy after important features such as Functional Assessment, ADL, Memory Complaints, MMSE, Behavioral Problems had been identified using Extra tree classifier. Various machine learning and deep learning models, like Random Forest (RF), Support Vector Machine (SVM), Convolutional Neural Networks (CNN) and XGBoost, were used and analyzed. A DeepLightBoost, a novel hybrid model, was also constructed through integrating the speeding up-based LightGBM with the learning-based Multi-Layer Perceptron (MLP). The model DeepLightBoost performed better with an accuracy of 95.58% in precision, recall, F1-score, and MCC. The findings underlie a great promise of hybrid paradigms for the reduction of false positive AD diagnosis, aid in early diagnoses, and delivery of personal based medicine plans.