Gait analysis has important value in detecting and warning Alzheimer's disease.Some studies suggest that gait disorders in individuals with cognitive impairments occur earlier than neuropsychological abnormalities.By analyzing gait to determine whether there is a mild cognitive impairment, the risk of developing Alzheimer's disease can be detected earlier.Early gait disorders require the use of dual task walking to be detected, while ordinary single task walking is masked by cognitive reserve, and its mechanism still needs further research.In the field of traditional medicine, medical staff manually observe, record, and analyze gait, which takes a long time, incurs high labor costs, and cannot be performed simultaneously on a large scale.The 3D gait analysis system and wearable gait analysis system have high accuracy, but there are shortcomings such as complex operation and high dependence on equipment and professionals.The article proposes a gait analysis method based on a monocular camera, aiming to solve the problems of accuracy, portability, and efficiency of existing gait analysis methods.Machine learning of gait parameters obtained through gait analysis can determine whether a user has mild cognitive impairment and thus determining the risk of developing Alzheimer's disease.The article provides the overall process of the method and proposes optimization strategies.The experiment has shown that this testing method has low cost, strong convenience, good reliability and validity of the gait parameters obtained, and high accuracy in determining whether there is mild cognitive impairment.