Background <p>Delirium is a common complication among older hospitalized adults and associated with poor prognosis, prediction of its occurrence is especially important. Our study aimed to establish and validate a delirium prediction model for hospitalized older medical patients.</p> Methods <p>We conducted a prospective study of 1067 hospitalized patients aged ≥ 65 years in the geriatric ward of West China Hospital of Sichuan University from November 2020 to April 2022. A nomogram model was established based on logistic regression. Nomogram performance was internally validated with Bootstrap resampling. The final model was externally validated using 98 patients from the Chengdu Sixth People’s Hospital from June to November 2022. Discrimination, calibration and clinical utility of the prediction model were assessed using the C-statistic, calibration plot and decision curve analysis, respectively.</p> Results <p>A total of 1067 patients were in the modelling group, 70 of whom developed delirium. The final model included age, cognitive impairment, polypharmacy, sleep disturbance, activity of daily life, malnutrition, and neutrophil-lymphocyte ratio (NLR). The area under the curve (AUC) of the model was 0.934, with sensitivity and specificity of 97.1% and 77.3%. The AUC after bootstrap internal validation was 0.925. External validation on 98 patients resulted in an AUC of 0.842.</p> Conclusion <p>We developed and externally validated a delirium prediction model in hospitalized older medical patients that showed good discrimination and calibration. Its application may be useful to identify patients at high risk of delirium to develop target interventions strategies.</p>