An On-Device Explainable Random Forest Expert System for Multi-Condition Health Screening on Resource-Constrained Mobile Devices

Abstract

This paper presents a four-layer on-device expert system architecture for multi-condition health screening on resource-constrained Android devices. The system uses a 50-tree Random Forest (4.2 MB) achieving 94.4% accuracy across 41 conditions with 28 ms inference latency. A sub-millisecond explainability module generates matched-feature lists, missing-feature analysis, and coverage ratios 200-800x faster than LIME/SHAP. A 30-participant user study found explainability features increased perceived trust by 83% and comprehension by 87%. The architecture is domain-agnostic and transfers to activity recognition and smart-home anomaly detection. Code and dataset: https://github.com/gebriye1986/Explain-Health