dsLassoCov: a federated Lasso approach incorporating covariate control

2026-04-22 · Source: Machine learning : nature.com subject feeds · Field: Science & Research — Artificial Intelligence & Machine Learning, Health & Medical Research, Mathematics & Computational Sciences · Depth: Expert, medium

Summary

dsLassoCov is a novel machine learning approach published on April 22, 2026, designed to address the challenge of covariate control in federated learning, particularly for high-dimensional biomedical data. Federated learning enables privacy-preserving model training across distributed datasets, but conventional covariate control methods often incur substantial communication costs. dsLassoCov offers an efficient solution, supporting the identification of biomarker candidates while accounting for confounding effects. Using simulated data, the method demonstrated effective management of confounding. In a real-world application, dsLassoCov successfully replicated a large-scale exposome analysis across six distinct databases, yielding results consistent with prior studies. This approach aims to accelerate federated learning applications in large-scale biomedical research by mitigating a key technical hurdle.

Key takeaway

For AI Scientists and Data Scientists working with sensitive, distributed biomedical datasets, dsLassoCov offers a practical solution for integrating covariate control into federated learning workflows. You should consider implementing dsLassoCov to efficiently manage confounding effects and accelerate biomarker discovery or large-scale exposome analyses without centralizing data, thereby adhering to data governance and privacy regulations.

Key insights

dsLassoCov efficiently controls covariates in federated learning for high-dimensional biomedical data, enabling privacy-preserving analysis.

Principles

• Federated learning preserves data privacy.
• Covariate control is crucial in biomedical analysis.
• Communication costs are a key challenge in federated learning.

Method

dsLassoCov is a federated Lasso approach that integrates covariate control, allowing efficient, privacy-preserving training of machine learning models on geographically distributed, high-dimensional datasets.

In practice

• Identify biomarker candidates with confounding control.
• Replicate large-scale exposome analyses.
• Train models on distributed biomedical datasets.

Topics

• dsLassoCov
• Federated Learning
• Covariate Control
• Biomedical Research
• Exposome Analysis

Code references

• transbioZI/dsMTLClient
• transbioZI/dsMTLBase

Best for: AI Scientist, Research Scientist, Data Scientist

Related on AIssential

• Federated orthogonal learning for detection of liver lesions from multi-phase contrast-enhanced CT images — FedOG enhances federated learning for medical image segmentation by using orthogonal gradient decomposition to counter data heterogeneity and incomplete phases.
• Anti-causal domain generalization: Leveraging unlabeled data — In anti-causal settings, utilizing unlabeled data from multiple environments can improve domain generalization by penalizing model sensitivity to covariate shifts.
• FedSIR: Spectral Client Identification and Relabeling for Federated Learning with Noisy Labels — FedSIR uses spectral analysis of client features to identify and relabel noisy data in federated learning.
• DeFed-GMM-DaDiL: A Decentralized Federated Framework for Domain Adaptation — DeFed-GMM-DaDiL enables decentralized multi-source domain adaptation using federated GMMs and Wasserstein barycenters.
• Comparing decentralized machine learning and AI clinical models to local and centralized alternatives: a systematic review — Decentralized learning offers clinically acceptable performance in healthcare, especially for privacy-constrained contexts, outperforming local learning.
• Exploiting Correlations in Federated Learning: Opportunities and Practical Limitations — Federated learning compression schemes exploit structural, temporal, or spatial correlations, which vary significantly by task and model.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.