Synthetic Multimodal Tear Biomarker Dataset for Ocular and Systemic Risk

Presenting a clinically constrained synthetic dataset for machine learning-based early risk prediction of ocular and systemic pathologies. It integrates tear biochemical biomarkers, inflammatory cytokines, physiological indicators, lifestyle parameters, and engineered diagnostic indices, with all values generated within physiologically reported ranges from literature.

Use Cases

• Predict the Composite Risk Score using tear biomarkers like osmolarity, electrolytes, and inflammatory cytokines such as IL-6 and TNF-α.
• Analyze correlations between lifestyle parameters like screen time and sleep duration with physiological indicators such as blink rate and redness score.
• Benchmark explainable AI models on engineered diagnostic indices like the Inflammation Index and Tear Film Dysfunction Index.
• Study the relationship between tear biochemical biomarkers (total protein, lactoferrin, lysozyme) and the Composite Risk Score for systemic disease risk assessment.

Strengths

• Data values are constrained within physiologically and pathologically reported ranges derived from peer-reviewed literature.
• Integrates multiple data modalities including biochemical, physiological, lifestyle, and engineered diagnostic indices.
• Designed specifically for research in non-invasive tear-based diagnostics and explainable AI analysis.

Limitations

• Dataset is entirely synthetic, containing no real patient data, which may limit real-world applicability.
• Unknown sample size (row count) and specific feature count (columns), making scale and dimensionality unclear.
• Potential bias from the literature-derived ranges used for data generation, which may not capture full population variability.

Provenance

Source

Harvard Dataverse, authored by N, AMRUTHA.

Collection Method

Synthetically generated based on ranges from peer-reviewed ophthalmology, tear proteomics, and systemic disease literature.

Freshness

Last updated on 2026-02-20.