Alvaro Javier Vargas Guerrero

Alvaro is a researcher specializing in Federated Learning, a paradigm that enables collaborative artificial intelligence model development across institutions operating under strict data protection regulations. In this framework, each institution trains a local model on its private data, which is then securely transmitted to a federated server. The server aggregates these models into a global model that captures shared knowledge while preserving the privacy of the underlying datasets, redistributing improved versions back to each institution until convergence is achieved.

Working independently, his current research applies Federated Learning to the healthcare sector, with a focus on developing early detection models for neurological diseases such as Alzheimer’s disease and Multiple Sclerosis. By leveraging limited MRI data distributed across multiple hospitals, Federated Learning allows for more robust model training without violating GDPR constraints. This approach helps mitigate data scarcity while maintaining the privacy and autonomy of each participating institution.

Recently, Alvaro has expanded his work to incorporate causality into Federated Learning. While traditional models often rely on correlations, causal inference aims to uncover the underlying mechanisms that truly drive disease progression. By integrating techniques such as structural causal models and counterfactual reasoning into federated workflows, his research seeks to identify invariant disease-related features that remain stable across hospitals, reducing biases caused by scanner differences or demographic disparities. The goal is to build privacy-preserving models that not only predict conditions accurately but also support causally grounded clinical insights and more trustworthy decision-making.