Cataract, Cerebral Small Vessel Disease, Cerebrovascular Stenosis, Diabetic Eye Diseases, Vitreous Hemorrhage
Conditions
Brief summary
This project integrates multimodal eye data (CFP, OCT, OCTA) from 50,000 cases, brain MR data from 150,000 cases, and ICD diagnoses, medical orders, and test results from an eye-brain paired cohort of 8,000 cases to construct an eye-brain cross-modal mapping model and an eye-brain-systemic disease association model. It aims to clarify the quantitative associations between multimodal ocular features and brain structural and vascular characteristics as well as systemic disease ICD diagnoses, thereby uncovering the cross-organ and cross-modal linkage mechanisms between the eye and brain. Ultimately, it seeks to achieve mapping and prediction of brain imaging features based on ocular data, enable ocular-based diagnosis and prediction of various diseases, and contribute significantly to early disease screening, risk stratification, and optimization of clinical diagnosis and treatment.
Interventions
OTHERNo interventions were involved in this study.
No interventions were involved in this study.
Sponsors
Beijing Friendship Hospital
Study design
Observational model
COHORT
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
ALL
Healthy volunteers
Yes
Inclusion criteria
* Eye data: Images must meet quality standards (no significant artifacts, key structures clearly visible), with complete examination records (including examination date and device model). * Brain data: MR images must be free of motion artifacts, with complete sequences (at least including T1-weighted imaging and MRA), and reports must confirm no technical issues. * Eye-brain paired cohort: Must contain at least one type of eye image (including CFP, OCT, or OCTA) along with brain MR data, and be linkable to a unique identifier and clinical information in the database * Clinical data: Clinical information (such as core biochemical indicators, demographic information, and necessary questionnaire items) must have a missing rate ≤30%, with standardized and clear ICD diagnostic codes, complete laboratory test data, and medical order information (including medication or follow-up recommendations) * Prior to initial inclusion in the study, participants must have no history of severe organic diseases (including non-curable or end-stage malignancies, severe heart failure (NYHA Class III-IV), end-stage renal disease (CKD Stage 5), decompensated cirrhosis, or significant functional impairment due to severe cerebrovascular disease sequelae, etc.).
Exclusion criteria
* Ocular examinations in which key features cannot be identified due to conditions such as cataracts or vitreous hemorrhage brain MR images deemed unacceptable in quality due to factors such as metal implants * Clinical information (including core biochemical indicators, demographic details, essential questionnaire items, etc.) with a missing rate exceeding 30% * Individuals who are pregnant or lactating, those with ocular trauma, congenital ocular malformations, or severe organic diseases.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Performance of Systemic Disease Prediction Models | From the time of model construction and consistency alignment during the study period to the end of the 12-month follow-up after study initiation. | Age- and sex-stratified Area Under the Receiver Operating Characteristic Curve (AUC) |
| Model generalisation capability. | From the completion of model construction and consistency calibration during the study period to the end of the 12-month follow-up after study initiation. | Evaluated by Cohen's kappa coefficient across external independent cohorts. |
| Eye-Brain Mapping Model Performance. | From the completion of model construction and consistency calibration during the study period to the end of the 12-month follow-up after study initiation. | Measured by correlation coefficient between retinal imaging features and brain neuroimaging metrics. |
Outcome results
None listed