Key factors influencing osteoporosis development in frail elderly patients with cerebral infarction include age, TFI score, GNRI, Barthel Index, and SAS score, and a nomogram built from these factors demonstrates predictive value, high calibration accuracy, and clinical utility.
Among 120 elderly frail patients with cerebral infarction, 41 developed osteoporosis and 79 did not.
Univariate comparison showed significant differences between the osteoporosis and non-osteoporosis groups on eight variables.
Three machine learning methods (LASSO, XGBoost, Random Forest) with overlap-covering screening converged on five common risk factors.
In logistic regression, age, TFI score, and SAS score were risk factors for osteoporosis, while GNRI and Barthel Index were protective factors.
The logistic model showed acceptable goodness of fit.
The nomogram built from the five factors showed high discrimination.
Calibration and decision curve analyses indicated good explanatory power, calibration, and net clinical benefit.
This study looked at 120 elderly patients who were both frail and had suffered a cerebral infarction (a type of stroke), to find out which factors were linked to having osteoporosis. Forty-one patients had osteoporosis and 79 did not. Using three machine learning methods to pick out the most consistent factors, the researchers identified five: age, frailty score (TFI), nutritional risk (GNRI), daily-living ability (Barthel Index), and anxiety score (SAS). Older age, greater frailty, and higher anxiety were associated with higher odds of osteoporosis, while better nutritional status and better ability to perform daily activities were associated with lower odds. The researchers combined these five factors into a nomogram, a scoring tool for estimating individual risk. It separated patients with and without osteoporosis very well in their data (AUC 0.955), and calibration and decision-curve analyses suggested it could be clinically useful. This research suggests that assessing frailty, nutrition, function, and anxiety in stroke patients could help flag those at higher risk of osteoporosis. However, the study was small, retrospective, and from a single hospital, and it was validated only internally, so the tool would need testing in other groups of patients before being relied on more widely.
Liu Y, Wen J, Chen Y, Li H, Xie P. (2026). A predictive model for osteoporosis in frail elderly patients with cerebral infarction: an integrated analysis based on machine learning and care-sensitive indicators.. Frontiers in endocrinology. https://doi.org/10.3389/fendo.2026.1771964