Related Experiment Video
Updated: Aug 13, 2026
09:05
In Vitro and In Vivo Models to Study Corneal Endothelial-mesenchymal Transition
Published on: August 20, 2016
Stability selection machine learning identifies biomechanical markers for subclinical idiopathic corneal endothelial
Chengjie Feng1, Miaomiao Chi1, Shaofeng Gu1
1Department of Ophthalmology, Peking University Third Hospital, Beijing, China.
Frontiers in Medicine
|August 12, 2026
Summary
Early detection of corneal endothelial compromise is challenging. Corneal biomechanics, particularly A2L and cTBI, show promise as noninvasive markers for identifying subclinical endothelial impairment.
Area of Science:
- Ophthalmology
- Biomedical Engineering
- Corneal Science
Background:
- Idiopathic corneal endothelial compromise is a common, underdiagnosed pre-cataract finding.
- Early detection remains a significant clinical challenge.
- Understanding the link between corneal biomechanics and endothelial status is crucial.
Purpose of the Study:
- Investigate the association between corneal biomechanical parameters and endothelial status.
- Identify potential biomechanical markers for early endothelial impairment.
- Explore biomechanical manifestations of idiopathic endothelial compromise.
Main Methods:
- Cross-sectional study of cataract patients using Corneal Visualization Scheimpflug Technology (Corvis ST) tonometry and specular microscopy.
- Employed stability selection, a machine learning method, to identify robust biomechanical predictors.
- Accounted for central corneal thickness (CCT) using stratified analysis and propensity score matching (PSM).
Main Results:
- Five biomechanical parameters (SSI, HCR, A2L, A1T, cTBI) differed significantly between normal and abnormal endothelium groups, even after PSM for CCT.
- A2L was identified as an independent protective factor (OR: 0.202).
- A2L and cTBI emerged as the most robust markers; a novel DASC model achieved an AUC of 0.711 for detecting endothelial abnormalities.
Conclusions:
- Corneal biomechanics reflect early endothelial alterations, indicating reduced corneal stiffness and impaired deformation.
- A2L, cTBI, and the DASC model are promising noninvasive tools for risk stratification of endothelial compromise.
- Subclinical endothelial impairment can be detected through specific changes in corneal biomechanical responses.