Alzheimer's disease dementia (AD) is a debilitating and prevalent neurodegenerative disease in older adults globally. Cognitive impairment, a hallmark of AD, is assessed through verbal tests that require high specialization, and while accepted as screening tools for AD, general practitioners seldom use them. AD can be diagnosed with expensive, invasive neuroimaging and blood tests, but these are usually conducted when cognitive functioning is already severely impaired. Thus, finding a novel, non-invasive tool to detect and differentiate mild cognitive impairment (MCI) and AD is a prime public health interest. Self-figure drawings (a projective tool in which individuals are asked to draw a picture of themselves), are easy to administer and have been shown to differentiate between healthy and cognitively impaired individuals, including AD. Convolutional Neural Network (CNN) (a type of deep neural network, applied to analyze visual imagery) has advanced to assess health conditions using art products. Therefore, the proposed study suggests utilizing CNN-based methods to develop and test an application tailored to differentiate between drawings of individuals with MCI, AD, and healthy controls (HC) using 4,000 self-figure drawings. This