Trust Aware Federated Learning for Secure Bone Healing Stage Interpretation in e-Health

TL;DR

Experimental results demonstrate that adaptive trust management can improve both training stability and predictive performance by mitigating the negative effects of compromised clients while retaining robust detection capabilities.

Abstract

This paper presents a trust aware federated learning (FL) framework for interpreting bone healing stages using spectral features derived from frequency response data. The primary objective is to address the challenge posed by either unreliable or adversarial participants in distributed medical sensing environments. The framework employs a multi-layer perceptron model trained across simulated clients using the Flower FL framework. The proposed approach integrates an Adaptive Trust Score Scaling and Filtering (ATSSSF) mechanism with exponential moving average (EMA) smoothing to assess, validate and filter client contributions.Two trust score smoothing strategies have been investigated, one with a fixed factor and another that adapts according to trust score variability. Clients with low trust are excluded from aggregation and readmitted once their reliability improves, ensuring model integrity while maintaining inclusivity. Standard classification metrics have been used to compare the performance of ATSSSF with the baseline Federated Averaging strategy. Experimental results demonstrate that adaptive trust management can improve both training stability and predictive performance by mitigating the negative effects of compromised clients while retaining robust detection capabilities. The work establishes the feasibility for adaptive trust mechanisms in federated medical sensing and identifies extension to clinical cross silo aggregation as a future research direction.

Figures (6)

• Figure 1: FL workflow for bone healing stage prediction with ATSSSF.
• Figure 2: Variance of client trust scores across 500 communication rounds. Adaptive EMA demonstrates faster stabilization and lower long term volatility than both static ATSSSF and baseline FedAvg, indicating improved robustness in trust estimation.
• Figure 3: Evolution of average trust scores (solid lines) and omitted clients per round (dashed lines) across 500 communication rounds. Adaptive EMA achieves faster stabilization of trust values and a steady reduction in omitted clients compared to the static configuration and baseline FedAvg.
• Figure 4: Confusion matrix for the baseline FedAvg model. Misclassifications are more frequent across adjacent healing stages, particularly between Soft Callus and Early Mineralization, reflecting the sensitivity of standard aggregation to client variability.
• Figure 5: Confusion matrix after applying the ATSSSF mechanism with fixed parameters ($\alpha = 0.3$, $\tau = 0.75$). Trust based filtering reduces off diagonal errors and improves classification balance across all stages.