Robust and interpretable unit level causal inference in neural networks for pediatric myopia - PubMed
9 hours ago
- #neural networks
- #causal inference
- #pediatric myopia
- Proposes a causal inference framework integrated into neural networks for assessing individual feature influence on predictions.
- Utilizes a pediatric ophthalmology cohort of over 3000 children with longitudinal follow-up to estimate direct and indirect causal effects.
- Achieves good performance and identifies clinically plausible causal pathways in myopia progression.
- Includes refutation experiments confirming the robustness and reliability of causal effects.
- Model-agnostic approach suitable for digital health interventions requiring explainability.
- Advances transparent and reliable AI systems aligned with precision medicine and equitable healthcare goals.