Towards Transparent Rare Disease Monitoring: Federated Rule Extraction for Online Streaming NPC1 Data
Abstract
Abstract The early and accurate diagnosis of Niemann-Pick disease type C1 (NPC1) poses a fundamental challenge for machine learning: clinical datasets are exceptionally small, severely imbalanced, geographically fragmented, and subject to strict privacy regulations. Furthermore, existing black-box models are ill-suited to this regime, lacking the data efficiency and interpretability necessary for clinical validation. This paper presents OFEFS (Online Federated Evolutionary Fuzzy System), a novel framework that addresses these constraints by integrating federated learning, online incremental adaptation, and evolutionary fuzzy rule induction. OFEFS ensures data privacy by keeping patient records localized, while continuously updating a globally shared model without requiring full retraining. To handle the adverse data characteristics typical of rare diseases, the pipeline incorporates fuzzy-based feature selection and synthetic oversampling to navigate extreme dimensionality and class imbalance. Experimental validation on a clinical cohort of NMR urine spectra demonstrates that the global federated model achieves a robust equilibrium between diagnostic sensitivity and clinical reliability, autonomously recovering rules that align with established biochemical pathways. Crucially, OFEFS synthesizes distributed knowledge into compact, human-readable linguistic rules (IF-THEN), providing clinicians with a transparent, trustworthy AI tool that preserves patient privacy and adapts…
How strong is the evidence?
Evidence Hub reads this as a cohort study, done in people. Its evidence level is 3 of 5 (Moderate · trial / cohort / secondary analysis). Trials without a large randomized design, cohorts and secondary analyses give moderate evidence.
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.