Swiss Data Science Center

Snow avalanches are among the deadliest natural hazards in mountainous regions. Yet avalanche activity is often still documented manually, and accurate avalanche release times are mostly missing. Automated monitoring systems equipped with seismic and infrasound sensors, combined with detection algorithms, could help record avalanche occurrences and provide accurate data on release time, size, and type. This comprehensive data on avalanche activity is indispensable for improving and validating avalanche forecasts and for implementing mitigation measures. At the Vallée de la Sionne (VDLS) test site in Switzerland, a combination of avalanche monitoring systems has been deployed for over two decades, including radars, cameras, seismic and infrasound stations. Additionally, avalanche researchers have manually documented and verified most avalanche events over the past 14 winter seasons to compile a unique avalanche catalogue.To facilitate and automate avalanche detection, we aimed to implement two deep learning-based methods that scan continuous seismic and infrasound data separately in (near) real-time to detect and classify avalanche signals. Therefore, we leveraged the large volume of continuous data collected every winter at VDLS by adopting concepts from recent, powerful language models. Specifically, we pre-trained transformer networks in a self-supervised manner (i.e. without using expert labels) on a wide variety of signals mined from continuous seismic and infrasound data streams. The models receive fixed-length waveforms as input, partition them into sequential patches and compute patch-wise spectrograms. By training the models to reconstruct a portion of randomly masked patches, they learn to extract meaningful representations from the data, achieving silhouette scores of up to 0.6. This indicates good separability between avalanche and non-avalanche signals. Thus, these representations can later be used to automatically detect avalanches by fine-tuning a classifier on top. Moreover, combining predictions from the seismic and infrasound models has the potential to further improve (near) real-time avalanche detection.

Weather foundation models are increasingly expected to operate under heterogeneous and imperfect observation settings while remaining computationally scalable. Building on the Earth System Foundation Model (ESFM) setting for heterogeneous data integration, we explore how Mixture-of-Experts (MoE) can support robust and efficient learning in multi-modal weather foundation models.We introduce ESFM-MoE, an exploratory direction that combines conditional computation with climate-semantic routing, a routing principle that encourages expert specialization aligned with meaningful geophysical structure, rather than treating expert selection as a purely generic scaling mechanism. The motivation is that Earth-system data exhibit strong spatial organization, regime-like variability, and modality-dependent uncertainties; MoE offers a natural way to allocate capacity adaptively and promote structured specialization under such heterogeneity.In this work, we discuss the design space and practical considerations of integrating MoE into Earth-system foundation models, focusing on how routing objectives and inductive biases can shape expert behavior and improve utilization. We highlight potential benefits for robustness to missing observations, scalable training and inference, and outline promising directions for climate-aware expert specialization in next-generation weather foundation models.

Catastrophic mass movements threaten mountain communities worldwide. Rockfalls, avalanches, debris flows and sediment pulses in rivers are common geomorphological processes but can destroy homes and infrastructure with little warning. Population pressure, thawing permafrost and other climatic affects will likely exacerbate this threat in the near future requiring new risk management strategies and monitoring tools.In recent years, seismology has emerged as an efficient observational method to capture rapid mass movements and study their dynamics as well as variations in event activity. Multi-million cubic meter rock-ice avalanches like the 2025 event destroying parts of the village of Blatten, Switzerland, are often detected by national seismic networks primarily designed to monitor earthquake activity. Smaller events like rockfalls and debris flows require denser seismic networks with station spacing of a few kilometres or less. Nevertheless, their seismic signature is usually clear when seismic stations are close enough.The straightforward detection of mass movements using seismic instrumentation has motivated new monitoring approaches. However, the challenge remains to automatically identify the seismic mass movement signature in continuous data streams given a wealth of other signals like anthropogenic noise and earthquakes, which are recorded at the same time and may mask the sought-after mass movement signals. Recent applications of machine learning algorithms have provided promising first results and allowed for mass movement detection in cases where empirical threshold-based triggering rules yield impermissible amounts of false positives.Here we present a new approach to detect mass movements signals in continuous seismic catalogues. To tackle the challenge of algorithm transferability between sites with different seismic background noise we treat mass movement signals as anomalies given their catastrophic nature and rare occurrence. We use the isolation forest algorithm to quantify the degree of anomaly (‘anomaly score’) associated with any recorded signal. Using data from polar fjord systems, our results show that anomaly detection can efficiently reduce continuous seismic data sets to a handful of signals, which are likely related to rock avalanches and glacier break-off events. On smaller scales, anomaly scores can be processed to identify general characteristics of debris flow seismograms recorded near active torrents. The anomaly score approach thus facilitates systematically searching for large-scale mass movement seismograms in earthquake monitoring data and may be a stepping stone for flexible and transferable detection algorithms for monitoring and warning purposes.

<strong class="journal-contentHeaderColor">Abstract.</strong> Catastrophic mass movements, such as rock avalanches, glacier collapses, and destructive debris flows, are typically rare events. Their detection is consequently challenging as annotated and verified events used as training data for instrumentation and algorithm tuning are absent or limited. In this work, we explore seismic mass-movement data through the lens of anomaly detection. The idea is to screen out segments of the data that are unlikely to contain mass movements by focusing only on anomalous signals, thereby reducing the number of signals to be studied, making downstream tasks such as expert labeling and clustering of events easier. To extract anomalous signals, we design a triggering algorithm using an anomaly score computed from an isolation forest obtained from sliding windows taken from the continuous data. The extracted signals are subjected to expert labeling and/or further analyzed by dynamic time warping, a popular technique used to evaluate the dissimilarity between different types of signals. We illustrate our approach by (a) mining for seismic signals of hazardous debris-flows in Switzerland's Illgraben catchment and (b) labeling of seismic mass movement data obtained from a Greenland seismometer network.

Nitrous oxide (N 2 O) emissions from cropland soils are a significant source of greenhouse gases in Sub-Saharan Africa (SSA), yet fluxes and drivers remain poorly quantified. This review synthesizes findings from 39 field-based studies across Sub-Saharan Africa to assess current knowledge, key drivers, and research gaps. Additionally, this review offers a focused analysis on the effect of nitrogen fertilizer application on N 2 O emissions. The main finding is that emissions were strongly influenced by nitrogen input rates, precipitation, cropping systems, soil types, and measurement durations. Emission factors often diverge from IPCC Tier 1 defaults, suggesting the need for context-specific values. However, research is unevenly distributed: Studies from Kenya dominate the literature, while large areas, especially in Central Africa, lack data. Methodological inconsistencies further limit comparability across studies. We identify priorities for future research, including geographic expansion, standardized protocols, and improved integration of climate and agronomic variables. Strengthening the regional evidence base is essential for accurate inventories and for developing climate-smart agricultural policies tailored to SSA’s diverse agroecosystems.

BackgroundNovel therapeutics for Friedreich ataxia employ diverse strategies to increase frataxin protein levels, and a better understanding of the relation to clinical outcomes could strengthen their use as pharmacodynamic markers, and potentially as surrogate endpoint in therapeutic development. An elaborate modelling framework was developed to evaluate the suitability of frataxin as a biomarker across assays, tissues and disease stages.MethodsFrataxin levels generated previously through two distinct assay platforms and from two separate clinical cohorts: whole blood frataxin was measured by a lateral-flow immunoassay (LF cohort), and a triple-quadrupole LC-MS/MS method (TQ cohort), which enables separate quantification of mature frataxin (FXN-M) and erythrocyte-specific frataxin (FXN-E). Results were compared descriptively with control and heterozygous carriers, and several distinct modelling strategies were employed to correlate them with clinical function.ResultsBoth cohorts represented the relevant disease spectrum, with minor differences in both genetic and clinical severity, which correlated with frataxin levels. Heterozygous carriers showed intermediate levels. Modelling confirmed the predictive value of frataxin across multiple clinical assessments, such as age of symptom onset, age at loss of ambulation and long-term progression. GAA1, the shorter repeat expansion, was confirmed as the dominant predictor of frataxin itself, and, in most situations, clinical function.Discussion and conclusionAlthough isoform biology and tissue-specific expression remain important considerations, peripheral frataxin quantification provides biologically grounded measure of the pathophysiology and disease progression, with strong potential for application in therapeutic trials. Frataxin is a valid clinical biomarker, and our findings support advancing its candidacy as a surrogate endpoint in Friedreich ataxia.

Abstract Unintended policy consequences in higher education may arise when reforms targeting specific student groups interact with stratified institutional landscapes, with potential spillover effects on structurally disadvantaged domestic populations. This article examines a policy reform aimed at international students—specifically, the 2014 tightening of admission criteria at a Swiss elite STEM higher education institution (HEI)—and its broader, unforeseen impact on first-generation domestic students. Using administrative data and a quasi-experimental design, we find lower attrition among international students after the policy change and no observable change in Swiss students’ graduation rates. At the same time, enrolment of Swiss students declines—most notably among first-generation students—with a compensating increase in enrolment at less selective institutions, concentrated in fields that overlap with the elite programmes affected by the policy. These patterns suggest that the reform altered both formal access and perceptions of attainability, risk and institutional fit, aligning with arguments of constrained educational choice. The findings highlight redistributive costs of regulating international student inflow—shifting opportunities away from first-generation domestic students—and call for system-level governance that anticipates indirect effects on equity.

Switzerland's Electronic Patient Record (EPR) system faces a critical adoption crisis: less than 2% of the population has opened an account despite years of investment. To understand this failure from the citizen's perspective, we launched the first "expedition" in which 120 volunteers ("salutonauts") systematically tested Switzerland's current digital health infrastructure by opening EPRs, requesting their health data and documenting their experiences. This represents the first real-world, citizen-led assessment of Switzerland's health data infrastructure. Our findings reveal three fundamental insights. First, citizens appreciate the concept of personally controlled health records governed by a trustworthy national framework. Second, the current EPR implementation is both impractical and economically untenable under existing conditions: opening an EPR nationwide would cost CHF 400 million (calculated from required man-hours alone), while uploading historic documents would cost CHF 1 billion, and obtaining insurance claims data would cost over CHF 5 billion in lost productivity. Third, the recently announced draft of the new Electronic Health Dossier Law (E-GD) comes at a critical moment, since evidence-based redesign of the EPR system is essential. 95% of participants valued comprehensive medical checkups producing structured data, demonstrating citizens' appetite for high-quality digital health records when they provide tangible benefits. Our expedition methodology, which is agile, citizen-centred and involves real-world testing, offers a replicable model for evaluating digital health initiatives before massive implementation costs are incurred. With the E-GD revision underway, Switzerland has a unique opportunity to develop an EPR system that serves patients and physicians above all. Successful implementation requires EPR opening at birth by default (with the ability to opt-out), the AI-powered automation of data collection and uploading, seamless healthcare professional interfaces, and using the EPR as the billing data gateway. The alternative - i.e. continuing the current approach - virtually guarantees failure based on economic factors alone.

International trade agreements are major avenues for facilitating global economic growth, ensuring smooth supply chain management across the globe, and coordination of international relations among other social and economic activities among nations. However, such agreements and activities thereof are inevitably prone to disagreements and violations of the guiding principles. Multilateral trade agreements, Bilateral Investment Treaties, and even Regional Trade Agreements often involve countries with economic, political, and legal backgrounds or trends that predisposes them to occasional disharmony with the provisions. It is the inevitability of the disharmony and disputes in the international trade agreements that moved commercial regulators like the World Trade Organization to come up with legal and alternative mechanisms for resolving the disputes. Nevertheless, the trade agreements are equally unique in terms of their provisions with potential impacts on arbitration as an established alternative mechanism for solving international investment disputes. This study sought to examine the possible economic and legal implications of international trade agreements, including bilateral investment treaties and regional trade agreements on arbitration of international investment disputes. It was informed by three specific objectives namely, analysis of the provisions and mechanisms of international trade agreements, assessing the effects of the international trade agreements on the legal framework governing investment dispute resolution, and examination of the economic consequences of investment dispute resolution under international trade agreements. Systematic review of relevant literature alongside case analysis were used as the research methods for the realization of the aforementioned objectives. Both bilateral investment treaties and regional trade agreements were establish to contain substantial and binding provisions most of which recognize and support arbitration in the resolution of their investment disputes. Multiple cases emanating from various trade agreements were established to have been settled through arbitral tribunals. Therefore, the study concludes by noting that different provisions of the international trade agreements have worked towards strengthening the position of arbitration as a mechanism of solving international investment disputes.