A5018969859- Machine Learning in Healthcare
- Genomics and Phylogenetic Studies
- COVID-19 diagnosis using AI
- Microbial Community Ecology and Physiology
- Soil Geostatistics and Mapping
- Biomedical Text Mining and Ontologies
- Tuberculosis Research and Epidemiology
- Multimodal Machine Learning Applications
- Occupational and environmental lung diseases
- Domain Adaptation and Few-Shot Learning
- Dermatology and Skin Diseases
- Artificial Intelligence in Healthcare and Education
- Sepsis Diagnosis and Treatment
- Psoriasis: Treatment and Pathogenesis
- French Language Learning Methods
- Generative Adversarial Networks and Image Synthesis
- Species Distribution and Climate Change
- Radiation Dose and Imaging
- Occupational exposure and asthma
- Computational Fluid Dynamics and Aerodynamics
- Numerical methods for differential equations
- Lung Cancer Diagnosis and Treatment
- Advanced Numerical Methods in Computational Mathematics
- Radiomics and Machine Learning in Medical Imaging
- Topic Modeling
IBM Research - Zurich
2018-2023
University of Turin
2014-2023
University of the Sacred Heart
2022
Università Cattolica del Sacro Cuore
2022
Istituti di Ricovero e Cura a Carattere Scientifico
2022
Sustainable Europe Research Institute
2021
IBM (United States)
2021
Musée Cantonal d'Archéologie et d'Histoire
2017
University of Verona
2005-2009
Institut de Mécanique des Fluides de Toulouse
2005
Abstract Background We studied cancer mortality and mesothelioma incidence in 974 male workers employed at least 6 months the Balangero mine (Italy), largest chrysotile Western Europe, active from 1917 to 1985. Methods Vital status as of 31 May 2013, causes deaths 1990 were ascertained. Past exposure asbestos by working area calendar period was estimated, based on historical data fibers concentrations. Individual cumulative assessed applying estimates job history cohort members. Standardized...
The biogeographic region of Argentinean Puna mainly extends at elevations higher than 3,000 m within the Andean Plateau and hosts diverse ecological communities highly adapted to extreme aridity low temperatures. Soils are typically poorly evolved geomorphology is shaped by drainage networks, resulting in vegetated endorheic basins hypersaline known as salar or salt flats. Local rely on soil fertility for agricultural practices pastures livestock rearing. From this perspective, investigating...
Videocapillaroscopy (VCP) can be used to explore microcirculatory modifications in skin diseases. Psoriasis presents a specifically altered capillaroscopic pattern with 'bushy' capillaries and disarranged microangioarchitecture. The aim of the present study is compare clinical psoriatic target lesion during topical therapy.Thirty patients chronic plaque psoriasis were included study. Clinical comparable lesions elbows analyzed different therapies (calcipotriol, betamethasone dipropionate...
Despite the progress in automatic detection of radiologic findings from chest X-ray (CXR) images recent years, a quantitative evaluation explainability these models is hampered by lack locally labeled datasets for different findings. With exception few expert-labeled small-scale specific findings, such as pneumonia and pneumothorax, most CXR deep learning to date are trained on global "weak" labels extracted text reports, or via joint image unstructured strategy. Inspired Visual Genome...
<i>Background:</i> Videocapillaroscopy (VCP) can be used to explore microcirculatory modifications in skin diseases. Psoriasis shows specific capillary alterations reflecting typical histopathological changes. <i>Objective:</i> To compare morphology, distribution and density psoriasis seborrheic dermatitis of the scalp for differential diagnosis. <i>Methods:</i> VCP was performed on histology-confirmed lesions 30 patients with chronic plaque psoriasis,...
Although long-term management of psoriasis is paramount, this approach challenging in clinical practice. In the recent PSO-LONG trial, a fixed-dose combination betamethasone dipropionate (BD) and calcipotriol (Cal) foam applied twice week on non-consecutive days for 52 weeks (proactive treatment) reduced risk relapse. However, role Cal/BD needs further clarifications. The ProActive Management (PAM) program, nationwide Italian project, aims at reaching consensus proactive psoriasis.A steering...
Abstract The largest chrysotile mine in Western Europe was active Balangero (Italy) from 1917 to 1990. We quantitatively assessed exposure asbestos the framework of a cohort study on mortality miners and millers. Using documents filed at Italian State Archive we reconstructed job-histories members. concentration fibres by work-area derived industrial hygiene surveys since 1968 monitoring programs 1975. Earlier exposures had been estimated based experimental reconstruction past working...
There are between 6,000 – 7,000 known rare diseases today. Identifying and diagnosing a patient with disease is time consuming, cumbersome, cost intensive requires resources generally available only at large hospital centers. Furthermore, most medical doctors, especially general practitioners, will likely see one if all. A cognitive assistant for differential diagnosis in provide the knowledge on all online, help create list of weighted access to evidence base which was created. The system...
Capturing actions during scientific experimentation is a cornerstone of reproducibility and collaborative research. While large multimodal models hold promise for automatic action (or activity) recognition, their ability to provide real-time...
Summary The somatic chromosome number of Isoëtes duriei, 2n=110, was counted in individuals collected on Mount Pisano, loc. Asciano (Pisa, Tuscany), a olive grove terrace at 60 m a. s. l. plant may be considered as decaploid (10x).
We develop various AI models to predict hospitalization on a large (over 110$k$) cohort of COVID-19 positive-tested US patients, sourced from March 2020 February 2021. Models range Random Forest Neural Network (NN) and Time Convolutional NN, where combination the data modalities (tabular time dependent) are performed at different stages (early vs. model fusion). Despite high unbalance, reach average precision 0.96-0.98 (0.75-0.85), recall (0.74-0.85), $F_1$-score 0.97-0.98 (0.79-0.83)...
Neural networks are prone to catastrophic forgetting when trained incrementally on different tasks. Popular incremental learning methods mitigate such by retaining a subset of previously seen samples and replaying them during the training subsequent However, this is not always possible, e.g., due data protection regulations. In restricted scenarios, one can employ generative models replay either artificial images or hidden features classifier. work, we propose Genifer (GENeratIve...