To develop and validate a risk prediction model for Alzheimer's disease (AD) in elderly patients. Methods: A total of 382 elderly patients who visited the Department of Neurology in our hospital from January 2020 to December 2022 were included in this study. The patients were divided into a model group (267 cases) and a validation group (115 cases) at a ratio of 7 ∶ 3. Demographics, clinical cognition indicators, imaging data and laboratory indicators were collected. The model group was further divided into AD and non-AD subgroups. General information, clinical cognitive indicators, imaging data and laboratory indicators were compared between the two subgroups. Variables were screened using LASSO regression, followed by multivariate logistic regression. A nomogram model was developed and validated according to the results of multivariate analysis. Results: In the model group, 67 out of 267 patients (25.09%) had AD. LASSO regression identified 10 potential predictors, including age, history of hypertension, family history of AD, RAVLT, FAQ, hippocampal sulcus ratio, lateral cerebral fissure ratio, apolipoprotein A1, apolipoprotein E, and C-reactive protein. Multivariate logistic regression analysis showed that age, history of hypertension, RAVLT, FAQ, hippocampal sulcus ratio, apolipoprotein A1, apolipoprotein E and C-reactive protein were independent predictors (P<0.05). The area under the curve (AUC) of the AD risk prediction model for the elderly based on the model group was 0.968 (95% CI 0.946～0.990). External validation using the validation group showed an AUC of 0.957 (95% CI 0.932～0.983), which closely aligned with the internal validation results. The calibration curve indicated a close fit to the standard curve. The decision curve analysis showed that the net benefit rate was greater than 0 when the probability threshold of the nomograph model for predicting AD in elderly neurology patients ranged from 0.15 to 0.88. Conclusion: The prevalence of AD in elderly neurology patients is influenced mainly by factors such as age, history of hypertension, and RAVLT. The nomogram model developed in this study exhibits high accuracy and discrimination in predicting the risk of AD.