EcoFair: Trustworthy and Energy-Aware Routing for Privacy-Preserving Vertically Partitioned Medical Inference

TL;DR

EcoFair is a novel framework for privacy-preserving dermatological diagnosis that reduces energy consumption by selectively activating complex models based on uncertainty and risk assessments.

Contribution

EcoFair introduces a lightweight routing mechanism that balances diagnostic accuracy, privacy, and energy efficiency in vertically partitioned medical inference.

Findings

EcoFair significantly reduces energy use on edge devices while maintaining accuracy.

Selective routing improves malignant case detection without retraining the global model.

The framework effectively combines uncertainty and risk scores for decision-making.

Abstract

Privacy-preserving medical inference must balance data locality, diagnostic reliability, and deployment efficiency. This paper presents EcoFair, a simulated vertically partitioned inference framework for dermatological diagnosis in which raw image and tabular data remain local and only modality-specific embeddings are transmitted for server-side multimodal fusion. EcoFair introduces a lightweight-first routing mechanism that selectively activates a heavier image encoder when local uncertainty or metadata-derived clinical risk indicates that additional computation is warranted. The routing decision combines predictive uncertainty, a safe--danger probability gap, and a tabular neurosymbolic risk score derived from patient age and lesion localisation. Experiments on three dermatology benchmarks show that EcoFair can substantially reduce edge-side inference energy in representative model pairings while remaining competitive in classification performance. The results further indicate that selective routing can improve subgroup-sensitive malignant-case behaviour in representative settings without modifying the global training objective. These findings position EcoFair as a practical framework for privacy-preserving and energy-aware medical inference under edge deployment constraints.