PFed-Signal: An ADR Prediction Model based on Federated Learning

TL;DR

PFed-Signal addresses biased ADR signals in FAERS by introducing a federated-learning driven data-cleaning pipeline and a Transformer-based ADR predictor trained on the cleaned data. The framework consists of Pfed-Split to generate ADR-based splits and ADR-Signal to identify and remove biased data via Euclidean-distance analysis between local and global models, followed by Transformer-based ADR prediction on the cleaned dataset. Empirical results on FAERS (2010–2024) show higher ROR/PRR values on cleaned data and superior accuracy, AUC, precision, F1, and recall compared with SVM, BCPNN, and RF baselines, with bias-detection accuracy reaching $0.97$. The work demonstrates that removing biased data improves signal strength and predictive performance, offering a more reliable tool for ADR surveillance and clinical decision-making, with potential for broader applicability to multi-source pharmacovigilance data.

Abstract

The adverse drug reactions (ADRs) predicted based on the biased records in FAERS (U.S. Food and Drug Administration Adverse Event Reporting System) may mislead diagnosis online. Generally, such problems are solved by optimizing reporting odds ratio (ROR) or proportional reporting ratio (PRR). However, these methods that rely on statistical methods cannot eliminate the biased data, leading to inaccurate signal prediction. In this paper, we propose PFed-signal, a federated learning-based signal prediction model of ADR, which utilizes the Euclidean distance to eliminate the biased data from FAERS, thereby improving the accuracy of ADR prediction. Specifically, we first propose Pfed-Split, a method to split the original dataset into a split dataset based on ADR. Then we propose ADR-signal, an ADR prediction model, including a biased data identification method based on federated learning and an ADR prediction model based on Transformer. The former identifies the biased data according to the Euclidean distance and generates a clean dataset by deleting the biased data. The latter is an ADR prediction model based on Transformer trained on the clean data set. The results show that the ROR and PRR on the clean dataset are better than those of the traditional methods. Furthermore, the accuracy rate, F1 score, recall rate and AUC of PFed-Signal are 0.887, 0.890, 0.913 and 0.957 respectively, which are higher than the baselines.

Figures (8)

• Figure 1: The framework of PFed-Signal: A comprehensive architecture for federated learning-based ADR prediction
• Figure 2: Workflow of Pfed-Split: Data preprocessing and segmentation for federated learning
• Figure 3: Distribution of eigenvalues before and after data normalization: Enhancing data quality
• Figure 4: Clean dataset based on federated learning: Identifying and removing biased data
• Figure 5: ADR prediction based on Transformer: Capturing complex signal correlations