Authors
Farinoosh Hosseini, Morahem Ashengroph, Loghman Alaei
Published in
Chemico-biological interactions. Pages 112412. Nov 02, 2026. Epub Nov 02, 2026.
Abstract
Acetylcholinesterase (AChE) inhibition remains an important therapeutic strategy for symptomatic treatment of Alzheimer's disease. Quantitative structure-activity relationship (QSAR) modeling may help identify molecular features associated with AChE inhibitory activity.
Forty-one carbamate derivatives with reported AChE inhibitory activities were analyzed using 1,420 DRAGON molecular descriptors. An 11-descriptor multiple linear regression model was developed using forward stepwise selection and evaluated through leave-one-out cross-validation, Y-randomization, applicability-domain analysis, nested cross-validation, and external validation.
The model showed good fit to the training data (R2 = 0.9207), adjusted (R2= 0.8907), (RMSE = 0.0766) and leave-one-out performance (Q2LOO = 0.8382), RMSECV = 0.1094). Y-randomization yielded an empirical probability of 0.0190 for obtaining a training (R2) at least as high as the observed value. However, supplementary nested cross-validation produced a mean out-of-fold (R2) of 0.178, while external validation yielded (R2ext = -0.0548), indicating limited generalizability. The selected descriptors represented electronic, topological, geometrical, and molecular-property characteristics.
The model identifies descriptor-activity associations within the studied carbamate series but has limited demonstrated external predictive capability. It should therefore be considered an exploratory QSAR model requiring further validation on larger, independent datasets before application to the design or prioritization of new AChE inhibitors.
PMID:
42971381
Bibliographic data and abstract were imported from PubMed on 03 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 16
- Comments 0