Alzheimer’s disease (AD) is a constant neurodegenerative condition that fundamentally influences more seasoned grown-ups, prompting mental deterioration and weakening day to day working. Early recognition of AD is imperative for successful treatment, possibly easing back the infection’s movement. Be that as it may, momentum analytic strategies, for example, neuroimaging and mental testing, are frequently intrusive, costly, and may not recognize the illness until it has essentially progressed. This paper proposes an advanced deep learning structure utilizing Convolutional neural networks (CNNs) to investigate versatility information caught through cell phone accelerometers. The structure is intended to recognize unpretentious changes in development designs related with beginning phase Alzheimer’s, separating it from typical maturing and different circumstances. Our methodology consolidates progressed preprocessing procedures and hyperparameter advancement to improve the model’s presentation. The proposed model accomplished an exactness of 94.7% in characterizing beginning phase Promotion, showing its true capacity as a harmless, practical device for early determination. This system can be incorporated into wearable advances, offering ceaseless observing and early intercession valuable open doors for in danger populaces, altogether affecting patient consideration and the executives.