5,555 research outputs found
Protocol for training MERGE: A federated multi-input neural network for COVID-19 prognosis
No full textFederated learning is a cooperative learning approach that has emerged as an effective way to address privacy concerns. Here, we present a protocol for training MERGE: a federated multi -input neural network (NN) for COVID-19 prognosis. We describe steps for collecting and preprocessing datasets. We then detail the process of training a multi -input NN. This protocol can be adapted for use with datasets containing both image- and table -based input sources. For complete details on the use and execution of this protocol, please refer to Casella et al.
Exact Monte Carlo simulation of diffusion and jump diffusion processes with financial applications
No full textUn episodio di cronaca nera nella Udine di fine Seicento. Olimpia, Livia, Brandimarte e gli usi della storia
No full textIl saggio analizza un episodio di violenza che coinvolge due donne nobili nella città di Udine alla fine del Seicento. Fonti documentarie diverse aiutano a ricostruire lo svolgersi dei fatti, l’azione giudiziaria avviata e la composizione extragiudiziale tra le parti, l’ambiente sociale e culturale della nobiltà cittadina. Una fonte letteraria ottocentesca consente inoltre di analizzare l’uso “moderno” che di questa storia viene fatto un secolo e mezzo più tardi
Characterization of the microbial diversity in the water fluxes of a wooded riparian strip set up for nitrogen removal
No full textThis research is part of a project aimed at verifying the potential of a specifically assessed wooded riparian zone in removing excess of combined nitrogen from the Zero river flow for the reduction of nutrient input into Venice Lagoon. Seasonal fluctuations of microbial populations in the water entering and leaving the wooded riparian strip were determined for at least two years. Combined approaches
involving cultivation, microscopic approaches and DNA bases techniques were adopted to characterize both culturable and total microbial community. Seven major bacterial lineages, namely Firmicutes, Gammaproteobacteria,Actinobacteria, Alphaproteobacteria, Betaproteobacteria, Flavobacteria, and
Sphingobacteria were present in water samples as revealed by 16S rDNA sequence analysis of the culturable fraction of the bacterial population.
Gammaproteobacteria were the most dominant both in spring and fall, although
distinct bacterial communities were clearly detectable for the two seasons.
However, while DGGE cluster analysis did not reveal significant differences
between irrigation and drainage ditches, a significant alteration was detected by
PCA based on 16S rDNA of the culturable fraction. Since the wooded riparian
strip was vigorously working in terms of N removing by plant uptake and
especially by microbial denitrification, as demonstrated by parallel studies
performed on the same experimental site (Gumiero et al., 2011), it is reasonable to
suppose that the culturable bacteria fraction is the one effectively carrying out the
required task. In other words, the wooded riparian buffer zone specifically assessed
for water remediation (nitrogen removal) is efficiently working as a result of the
special conditions there produced to support the work of specific microbial
populations. This is confirmed by the increase of metabolically active bacteria
detected at the drainage ditches.
Taken together, the overall results provide key indications for the management of a
phytoremediation sit
The Digital Twin of the Metropolitan Area of Milan: Quality Assessment of Aerial and Terrestrial Data
No full textDigital twins have emerged as a promising technology for city planning and management, paving the way for the development of smart cities. Milan, Italy, has set an impressive example by planning to create a digital twin of its entire Metropolitan Area, covering a vast expanse of 1500 km(2). In late 2020, a tender was issued to collect aerial nadir and oblique images, LiDAR, and terrestrial mobile mapping data. The project will generate advanced products such as true orthophoto, classified LiDAR point cloud, DTM, DSM models, MMS point clouds and spherical depth images, and a database of 22 urban objects. To ensure the accuracy and consistency of the datasets, complex GNSS and terrestrial LiDAR measurements have been included for ground control and quality checks. The surveying activities were completed, and the data were delivered in mid-2023. The paper provides an overview of the quality assessment of aerial and terrestrial data, describes the datasets, analyzes image resolution, and discusses the accuracy and precision of acquired dataset, LiDAR, and imagery
Preoperative predictors of sternotomy need in mediastinal goiter management
No full textBACKGROUND: The objective of this study was to identify the preoperative risk factors for patients in need of a sternotomy in the management of mediastinal goiters in order to provide better preoperative planning and patient consent.
METHODS: We analyzed 98 patients who underwent surgery for mediastinal goiters (goiters extending below the thoracic inlet > or =3 cm with the neck in hyperextension) between 1995 and 2008. Twelve (12.2%) of the patients required a sternotomy. The patients' features were analyzed by the surgical approach performed. Logistic regression analysis was used to study which variables were influencing the surgical strategy. The receiver operating characteristic (ROC) curves were designed when appropriate.
RESULTS: The analysis disclosed the following risk factors: radiologic extension of mediastinal goiters below the aortic arch (odds ratio [OR] = 32.87; 95% confidence interval [CI] = 4.04-267.12; p 160 months: OR = 22.8; 95% CI = 5.28-98.53; p < .0001).
CONCLUSIONS: Sternotomy need for mediastinal goiter removal can be predicted; in such cases surgeons should not hesitate to perform it for minimizing complications
Introducing “La fabrique du droit”. A Conversation with Bruno Latour
No full textBruno Latour talks with Paolo Landri about his book on the Conseil d'Etat (La Fabrique du droit). The conversation was held in 2006 at the time of the Italian translation of the book and illustrates the research project and the difficulties the author had in the field. At the same time, it clarifies the trajectories of Bruno Latour's work and theoretical framework of his program of study with respect to sociology, anthropology, and philosophy of law. The conversation helps to understand the open-ended character of Bruno Latour's research and reflection including STS as well as sociological, anthropological and philosophical themes
MERGE: A model for multi-input biomedical federated learning
No full text: Driven by the deep learning (DL) revolution, artificial intelligence (AI) has become a fundamental tool for many biomedical tasks, including analyzing and classifying diagnostic images. Imaging, however, is not the only source of information. Tabular data, such as personal and genomic data and blood test results, are routinely collected but rarely considered in DL pipelines. Nevertheless, DL requires large datasets that often must be pooled from different institutions, raising non-trivial privacy concerns. Federated learning (FL) is a cooperative learning paradigm that aims to address these issues by moving models instead of data across different institutions. Here, we present a federated multi-input architecture using images and tabular data as a methodology to enhance model performance while preserving data privacy. We evaluated it on two showcases: the prognosis of COVID-19 and patients' stratification in Alzheimer's disease, providing evidence of enhanced accuracy and F1 scores against single-input models and improved generalizability against non-federated models
Recensione a Bruno Centrone, Vita in comune. Il pitagorismo nel mondo antico, Carocci, Roma 2024, 292 pp.
No full text- …
