Archivio della ricerca - Fondazione Bruno Kessler
Not a member yet
21227 research outputs found
Sort by
The TouCAN Codebook: Detecting textual misunderstanding in doctor-patient communication with the philosophy of language tools
No full textEffective communication is widely recognized as a cornerstone of successful medical treatment, as extensively documented in prior research. The doctor-patient relationship relies on clear, accurate information exchange to ensure precise diagnoses, treatment adherence, and patient satisfaction. Yet misunderstandings can seriously undermine this process, creating barriers to optimal care and weakening the therapeutic alliance-a critical element of effective healthcare. Consequently, identifying, understanding, and addressing these misunderstandings is essential. This paper introduces a novel approach to detecting and analyzing such misunderstandings in clinical interactions by drawing on concepts from the philosophy of language. Specifically, we present the ToUCAN Codebook, a structured framework designed to systematically classify and examine instances of textual miscommunication in doctor-patient dialogues. By offering a clear methodology for identifying and preventing potential communication breakdowns, the ToUCAN Codebook contributes to improved healthcare outcomes and a deeper understanding of the dynamics that shape medical discourse
Separation Power of Equivariant Neural Networks
No full textThe separation power of a machine learning model refers to its ability to distinguish between different inputs and is often used as a proxy for its expressivity. Indeed, knowing the separation power of a family of models is a necessary condition to obtain fine-grained universality results. In this paper, we analyze the separation power of equivariant neural networks, such as convolutional and permutation-invariant networks. We first present a complete characterization of inputs indistinguishable by models derived by a given architecture. From this results, we derive how separability is influenced by hyperparameters and architectural choices—such as activation functions, depth, hidden layer width, and representation types. Notably, all non-polynomial activations, including ReLU and sigmoid, are equivalent in expressivity and reach maximum separation power. Depth improves separation power up to a threshold, after which further increases have no effect. Adding invariant features to hidden representations does not impact separation power. Finally, block decomposition of hidden representations affects separability, with minimal components forming a hierarchy in separation power that provides a straightforward method for comparing the separation power of models
Fully-Geometric Cross-Attention for Point Cloud Registration
No full textPoint cloud registration approaches often fail when the overlap between point clouds is low due to noisy point correspondences. This work introduces a novel cross-attention mechanism tailored for Transformer-based architectures that tackles this problem, by fusing information from coordinates and features at the super-point level between point clouds. This formulation has remained unexplored primarily because it must guarantee rotation and translation invariance since point clouds reside in different and independent reference frames. We integrate the Gromov-Wasserstein distance into the cross-attention formulation to jointly compute distances between points across different point clouds and account for their geometric structure. By doing so, points from two distinct point clouds can attend to each other under arbitrary rigid transformations. At the point level, we also devise a self-attention mechanism that aggregates the local geometric structure information into point features for fine matching. Our formulation boosts the number of inlier correspondences, thereby yielding more precise registration results compared to state-of-the-art approaches. We have conducted an extensive evaluation on 3DMatch, 3DLoMatch, KITTI, and 3DCSR datasets. Project page: https://github.com/twowwj/FLAT
Measurement of the associated production of a top-antitop-quark pair and a Higgs boson decaying into a bb pair in pp collisions at √s = 13 TeV using the ATLAS detector at the LHC
No full textReconsidering the Value of Multi-Religious Spaces Based on the Notion of Religious Cultural Heritage: Beyond a Purely Symbolic or Entirely Utilitarian Function
Get PDFSociological research increasingly examines the diversity of cultural and religious resources that various community groups contribute to urban spaces and the public sphere. A key focus within this field is the reinterpretation of shared religious and spiritual spaces as part of the tangible and intangible religious cultural heritage. Adopting a spatial perspective, this analysis focuses on the specific case of top-down multi-religious places. Through an exploration of representative examples, this article investigates the different typologies of these places—from complexes that host distinct spaces for different faiths or religions to interfaith chapels and prayer and meditation rooms located in non-religious settings—using the framework of religious cultural heritage. The central conceptual bases of this framework—namely, the historical and memorial value, aesthetic considerations, sacredness and social function—are discussed in terms of their partial and complex association with the qualities of these unconventional spaces. This article suggests that the significance of multi-religious places from the perspective of religious cultural heritage is greater when these places do not serve merely a symbolic function or a purely pragmatic one. This article emphasizes the significance of spatial elements shaped by architectural design and construction choices, which can play a crucial role in integrating multi-religious spaces into the collective memory and foster appreciation for unique forms of sacred beauty
AI for high-resolution climate data: downscaling climate projections and decadal predictions with a deep learning Latent Diffusion Model
No full textBeveled microneedles with channel for transdermal injection and sampling, fabricated with minimal steps and standard MEMS technology
No full textMicroneedles hold the potential for enabling shallow skin penetration applications where biomarkers are extracted from the interstitial fluid (ISF) and drugs are injected in a painless and effective manner. To this purpose, needles must have an inner channel. Channeled needles were demonstrated using custom silicon microtechnology, having several needle tip geometries. Nevertheless, all the proposed fabrication sequences are not compatible with mass production based on mature, standard microfabrication techniques. Furthermore, ISF extraction was also demonstrated with channeled needles but under poorly controlled conditions and over long periods of time, the latter being impractical for medical use. A range of factors may impede or slow ISF extraction that require controlled experiments. In this work we address the above tasks in terms of microfabrication sequence design, tip geometry design and experimental validation under controlled conditions. We report the development and fabrication of a silicon channeled microneedle array using conventional, industrial micromechanic processes. With only 2 lithography steps, a hypodermic needle tip profile is achieved. Using the fabricated microneedles, fluid extraction is experimented on chicken skin mockups. Extraction tests are carried out by inducing a controlled pressure gradient between the two ends of the microneedle channels, generated by loading the chip or by applying vacuum to the chip's backside. The extraction of more than 1 μL of fluid in 20 minutes is demonstrated with a maximum applied pressure gradient of 500 mbar. A correlation between the extraction rate efficiency and needles' density is observed, both for short and long extraction times. These results provide the first demonstration of in vitro interstitial fluid collection under controlled experimental conditions using silicon hollow microneedles fabricated with standard micro electro mechanical systems (MEMS) fabrication technology and minimal steps. Based on the obtained data, a comparison is drawn between pressure load and vacuum as drivers for ISF extraction, according to modelling and controlled experiments
Case-time series study on the short-term impact of meteorological factors on West Nile Virus incidence in Italy at the local administrative unit level, 2012 to 2021
No full textIntroduction: West Nile Virus (WNV) is a significant public health concern in southern Europe, with meteorological, climatic, and environmental factors playing a critical role in its transmission dynamics. This study aims to assess the short-term effects of meteorological variables on the incidence of WNV in five Italian regions in Northern Italy from 2012 to 2021. Methods: Linking epidemiological data from the national surveillance system and local meteorological data, we conducted a Case-Time Series analysis to examine the association between WNV incident cases and temperature, humidity, and precipitation recorded up to ten weeks before case occurrence at the local administrative unit level. We employed conditional quasi-Poisson regression and distributed lag non-linear models to explore delayed effects. Results: Our study analyzed 1110 autochthonous human cases of WNV. We found a positive association between WNV incidence and weekly mean temperature recorded between one to nine weeks before the diagnosis, with the highest effect at one week lag (IRR: 1.16; 95% CI 1.11-1.21). An increase in weekly precipitations between the sixth and ninth weeks before diagnosis was also positively associated with WNV incidence. Variations in minimum weekly humidity did not show a consistent impact. Conclusions: Our findings underscore the influence of temperature and, to a lesser extent, precipitation on WNV incidence in Northern Italy, highlighting the potential of climatic data in developing early warning systems for WNV surveillance and public health interventions
Unveiling the drivers of active participation in social media discourse
Get PDFThe emergence of new public forums in the form of online social media has introduced unprecedented challenges to public discourse, including polarization, misinformation, and the rise of echo chambers. Existing research has extensively examined these topics by focusing on the active actions performed by users, without accounting for the share of individuals who consume content without actively interacting with it. In contrast, this study incorporates passive consumption data to investigate the prevalence of active participation in online discourse. We introduce a metric to quantify the share of active engagement and analyze over 17 million pieces of content linked to a polarized Twitter debate to understand its relationship with several features of online environments, such as echo chambers, coordinated behavior, political bias, and source reliability. Our findings reveal a significant proportion of users who consume content without active interactions, underscoring the importance of considering also passive consumption proxies in the analysis of online debates. Furthermore, we found that increased active participation is primarily correlated with the presence of multimedia content and unreliable news sources, rather than with the ideological stance of the content producer, suggesting that active engagement is independent of echo chambers. Our work highlights the significance of passive consumption proxies for quantifying active engagement, which influences platform feed algorithms and, consequently, the development of online discussions. Moreover, it highlights the factors that may encourage active participation, which can be utilized to design more effective communication campaigns
A cost-effective approach to counterbalance the scarcity of medical datasets
No full textThis paper presents an innovative methodology for addressing the critical issue of data scarcity in clinical research, specifically within emergency departments. Inspired by the recent advancements in the generative abilities of Large Language Models (LLMs), we devised an automated approach based on LLMs to extend an existing publicly available English dataset to new languages. We constructed a pipeline of multiple automated components which first converts an existing annotated dataset from its complex standard format to a simpler inline annotated format, then generates inline annotations in the target language using LLMs, and finally converts the generated target language inline annotations to the dataset's standard format; a manual validation is envisaged for erroneous and missing annotations. By automating the translation and annotation transfer process, the method we propose significantly reduces the resource-intensive task of collecting data and manually annotating them, thus representing a crucial step toward bridging the gap between the need for clinical research and the availability of high-quality data