Abstract
This letter critiques the article by Xu et al in World Journal of Psychiatry, which developed a nomogram to predict cognitive impairment in elderly hypertensive patients using nutritional and biochemical parameters. While the model's use of variables like body mass index, albumin, hemoglobin, alkaline phosphatase, and mini-nutritional assessment scores is promising, we raise concerns about the small validation cohort size, potential reverse causality in cross-sectional data, insufficiently discussed mechanisms for alkaline phosphatase as a risk factor, omission of key cognitive predictors, and possible overfitting given high area under the curve values. We suggest external validation, longitudinal studies, and calibration metrics to enhance the model's robustness and clinical utility.