Signals, in order.

The research presents a machine learning framework for early CKD diagnosis using structured clinical data from the UCI Machine Learning Repository. Multiple algorithms—logistic regression, SVM, random forest, XGBoost, LightGBM, and CatBoost—were evaluated within a single preprocessing pipeline, achieving F1-scores between 0.98 and 0.99, with SVM attaining the highest mean F1 of 0.9870. Validation through label permutation confirmed that the models learned meaningful clinical correlations rather than random noise.

Source: OpenAlex

A study by researchers from EPFL, ETH Zurich and the University of Zurich demonstrates how high‑resolution satellite imagery, combined with text‑based data, can improve the assessment of damage, population movements and infrastructure conditions in conflict zones. The research, published in Nature, shows that fusing satellite observations with other sources can uncover otherwise undocumented violence and provide a more complete picture of conflict dynamics.

Source: EPFL News

ETH Zurich computer scientists led by Rasmus Kyng have developed a new solver for Laplacian equations that uses randomized Approximate Cholesky factorization, achieving nearly linear running time and up to five times faster performance than existing software. The prototype, published in the SIAM Journal on Scientific Computing, reliably solves large network problems—including those where previous solvers fail—by sampling a small set of connections to reduce computational cost while maintaining accuracy. This work demonstrates how theoretical breakthroughs can be translated into practical, hig…

Source: ETH Zurich News

EPFL researchers have created ultra‑thin silicon metasurfaces that can dynamically tune their interaction with mid‑infrared light in real time, achieving record optical performance. By using integrated micro‑heaters and ultrafast laser pulses, the devices can modulate light at kilohertz to nanosecond speeds without altering their physical structure. The platform is scalable, compatible with large‑scale semiconductor manufacturing, and could enable compact, low‑power free‑space optical communications and adaptive chemical sensing.

Source: EPFL News

ETH Zurich will host an IBM Quantum System Two, powered by an IBM Nighthawk processor, at the CSCS Swiss National Supercomputing Centre in Lugano. The installation, funded through an offset agreement between Lockheed Martin and armasuisse, will operate for three years and provide access to researchers and selected Swiss companies. The system aims to advance research in materials science, chemistry, and quantum computing, and to support training of future quantum talent.

Source: ETH Zurich News

The article opens with a retrospective on the 1910 book "Die Welt in hundert Jahren" and then discusses when media linguistics became an established discipline and which media uses were early research focuses. It then turns to the emerging area of human‑machine interaction, specifically written interaction with large language models such as ChatGPT, a topic gaining increasing importance in media linguistics. The final section outlines future research themes for media linguistics, its relationship to AI research, and the impending challenges posed by the development of strong AI or artificial …

Source: OpenAlex

This research presentation, delivered in Reading, UK, at the S2S2D Conference 2026, focuses on developing 500‑year multi‑member seasonal climate hindcasts using machine learning methods. It highlights methodological advances and potential applications in long‑term climate forecasting.

Source: OpenAlex

The study by Valerie Sticher and colleagues at ETH Zurich combines text-based casualty data with automatically analysed satellite imagery to map conflict damage in Ukraine and Myanmar. It shows that satellite data can reveal ongoing violence beyond initial massacres and that casualty figures alone miss displacement and differing impacts on cities. The research highlights the need for diverse training data and interdisciplinary collaboration to improve automated conflict monitoring.

Source: ETH Zurich News

EPFL researchers developed an AI-based method that can detect and classify microplastics as small as 5 µm in Lake Geneva and surrounding rivers. The technique revealed that concentrations of 1–100 µm particles may have been underestimated by more than 650‑fold, with six of 38 samples exceeding ecotoxicological thresholds. The method uses two algorithms to separate microplastics from biological material and classify them into six polymer categories.

Source: EPFL News

Linguist Giorgio Iemmolo argues that, despite rapid advances in AI translation tools, learning foreign languages remains essential because languages encode distinct cultural and argumentative structures. Research comparing English and German scientific texts shows that language choice shapes argument flow, voice, and reader engagement. The Language Center of UZH and ETH Zurich offers over 500 language courses annually, including new academic communication workshops for employees and doctoral students.

Source: ETH Zurich News