Authors
Jing Bai, Xiaobing Li, Xin Guo, Bo Hui, Yaqian Mu, Wanting Li, Lei Zhang, Yaowen Luo, Junkai Cheng, Juan Li, Lei Shang, Bo Wang, Yan Shi, Jingjing Cao, Gejuan Zhang, Rui Li, Ya Bai, Xuedong Liu
Published in
Parkinsonism & related disorders. Volume 154. Pages 109063. Nov 14, 2026. Epub Nov 14, 2026.
Abstract
To develop and validate machine learning-based predictive prognostic models for assisted walking in patients with Multiple System Atrophy (MSA).
This was a multicenter bidirectional cohort study involving 173 patients with MSA (training set: 122; validation set: 51) from three hospitals (May 2022-January 2025). Baseline data (demographics, clinical manifestations, imaging, neurofunctional assessments) were collected, with follow-up until assisted walking occurrence. We constructed Cox regression and five machine learning prognostic models, which were evaluated via C-index, AUC, calibration curves, DCA, and K-M curves (Epidata, SPSS 25.0, R v4.2.2).
68.97% of patients required assisted walking (median: 48 months). Patients with MSA-C exhibited higher assisted walking proportion (75% vs 60.27%, P < 0.05) and shorter median time (47 vs 49 months, P = 0.042) than patients with MSA-P. Cox and CoxBoost models (incorporating six predictors) had training/validation C-indexes of 0.878/0.749 and 0.875/0.748, respectively, with stable AUC and good calibration.
MSA-P and MSA-C have distinct clinical features. The six identified predictors influence assisted walking outcomes. With acceptable stability and accuracy, the Cox and CoxBoost models exhibit promising prognostic performance of patients with MSA.
PMID:
43063725
Bibliographic data and abstract were imported from PubMed on 21 Nov 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 5
- Comments 0