Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.
Villarrica or Rukapillan (35.9°S; 2,847 m a.s.l.) is one of the most active volcanoes in South America and is the highest-risk volcano in Chile. It has an open conduit with a persistent lava lake. On the 3 March 2015, Strombolian activity rapidly progressed into a 1.5-km-high lava fountain, erupting at least ∼ 2.4 × 106 m3 of tephra. Soon after, the activity returned to mild Strombolian “background” explosions, which lasted until early 2017. Understanding the pre-eruptive conditions of such paroxysmal events is fundamental for volcanic hazard assessment. We present major and trace element geochemistry for glass and crystalline phases of basaltic andesite paroxysm pyroclasts (52–56 wt.% SiO2), and for the subsequent Strombolian “background” activity through February 2017 (54–56 wt.% SiO2). The lava fountain source magma was initially stored in a deeper and hotter region (9.4–16.3 km; ca. 1140 °C) and was then resident in a shallow (≤ 0.8 km) storage zone pre-eruption. During storage, crystallising phases comprised plagioclase (An66–86), olivine (Fo75–78) and augite (En46–47). Equilibrium crystallisation occurred during upper-crustal magmatic ascent. During storage in the shallower region, magma reached H2O saturation, promoting volatile exsolution and over-pressurization, which triggered the eruption. In contrast, subsequent “background” explosions involving basaltic-andesite were sourced from a depth of ≤ 5.3 km (ca. 1110 °C). Pre-eruptive conditions for the 2015 lava fountain contrast with historical twentieth-century eruptions at Villarrica, which were likely driven by magma that underwent a longer period of mixing to feed both effusive and explosive activity. The rapid transition to lava-fountaining activity in 2015 represents a challenging condition in terms of volcano monitoring and eruption forecasting. However, our petrological study of the pyroclastic materials that erupted in 2015 offers significant insights into eruptive processes involving this type of eruption. This aids in deciphering the mechanisms behind sudden eruptions at open conduit systems.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.