68994 research outputs found
Sort by
Development and application of innovative sensor networks for the Active and Assisted living environment
No full textQuesta tesi di dottorato indaga lo sviluppo e l'applicazione di reti di sensori innovative progettate per migliorare gli ambienti di Active and Assisted Living (AAL), con un focus sul supporto alla popolazione anziana. Affrontando le principali sfide nella localizzazione indoor e nel riconoscimento delle attività umane (HAR), questo lavoro propone soluzioni non invasive e rispettose della privacy, con l'obiettivo di migliorare la qualità della vita degli anziani e ridurre il carico sui caregiver.
La ricerca introduce tre contributi distinti: un sistema di localizzazione basato sulla Direzione di Arrivo (DOA), che utilizza sensori robotici per un posizionamento preciso in ambienti interni; una tecnica di fusione multi-sensore che integra sensori a infrarossi passivi (PIR) e sensori a ultrasuoni su un robot mobile, permettendo una localizzazione robusta e la ricostruzione delle traiettorie; e un nuovo approccio al riconoscimento delle attività umane, che utilizza una rete di sensori multi-dominio combinando sensori a ultrasuoni e ambientali per una classificazione accurata delle attività, garantendo al contempo la privacy degli utenti.
La tesi pone l'accento sui compromessi nell'integrazione dei sensori, in particolare sulla complessità degli algoritmi di fusione, sull'affidabilità e sulla gestione dell'incertezza, analizzando le implicazioni per l'implementazione pratica. I risultati sperimentali dimostrano progressi significativi nelle prestazioni di localizzazione e riconoscimento delle attività. Tuttavia, permangono sfide nel mantenere l'accuratezza in ambienti variabili, garantire la scalabilità e affrontare scenari con più residenti. Allineando queste tecnologie con il loro attuale livello di maturità tecnologica (Technology Readiness Level, TRL) e identificando percorsi per ulteriori sviluppi, questo lavoro pone le basi per future ricerche e implementazioni pratiche, promuovendo un ecosistema di supporto che favorisce l'autonomia degli anziani e migliora l'efficienza dell'assistenza.This Ph.D. thesis investigates the development and application of innovative sensor networks designed to enhance Active and Assisted Living (AAL) environments, with a focus on supporting the aging population. By addressing key challenges in indoor localization and Human Activity Recognition (HAR), this work proposes non-invasive and privacy-preserving solutions that aim to improve the quality of life for elderly individuals while alleviating the burden on caregivers. The research introduces three distinct contributions: a Direction of Arrival (DOA)-based localization system employing robotic sensors for precise indoor positioning; a multi-sensor fusion technique integrating Passive Infrared (PIR) and ultrasonic sensors on a mobile robot to enable robust localization and trajectory reconstruction; and a novel HAR approach using a multi-domain sensor network, which combines ultrasonic and environmental sensors for accurate activity classification while ensuring user privacy. The thesis emphasizes the trade-offs in sensor integration, particularly the complexity of fusion algorithms, reliability, and uncertainty management, and explores their implications for practical deployment. Experimental results demonstrate significant advancements in localization and activity recognition performance. However, challenges remain in maintaining accuracy across variable environments, ensuring scalability, and addressing multi-resident scenarios. By aligning these technologies with their current Technology Readiness Level (TRL) and identifying pathways for further development, this work lays the foundation for future research and practical implementations, fostering a supportive ecosystem that promotes independent living for aging populations and enhances caregiving efficiency
Assessing hypoglycemia risk during hemodialysis using an explainable machine learning approach based on continuous glucose monitoring metrics
No full textContinuous glucose monitoring (CGM) can identify hypoglycemia in hemodialysis (HD) patients, who are at risk for this event. On the other hand, machine learning has remarkable value in CGM-based studies, but none of previous studies addressed the problem of hypoglycemia prediction in HD patients. Therefore, we conducted this study to setup different machine learning models based on CGM data (specifically, CGM metrics) to assess the risk of mild and severe hypoglycemia during HD sessions. We studied a cohort of twenty patients (11 with and 9 without diabetes) undergoing chronic HD. All patients underwent CGM for up to 2 weeks. We identified 92 HD sessions and related pre-HD sessions of 8 h length. HD sessions were used to identify mild (<70 mg/dL) and severe (<54 mg/dL) hypoglycemia, whereas pre-HD sessions were used to compute 48 CGM metrics. We then performed feature selection to identify the most relevant metrics for hypoglycemia prediction. The metrics performance was assessed with binary decision tree, k-nearest neighbors, penalized logistic regression, Naïve Bayes, random forest ensemble algorithm. We found that mild hypoglycemia was best predicted by six metrics (M-value100, TIR70-180, ADRR, MAGE-, MAG30, CONGA1h), whereas severe hypoglycemia by three metrics (TIR70-180, ADRR, CONGA1h). The best overall performance was achieved by the tree, showing area under receiver operating characteristic curve (AUC) equal to 68.2 % for prediction of mild hyperglycemia, and AUC equal to 81.2 % for severe hyperglycemia. Notably, individual hypoglycemia risk assessment has potential to guide personalized HD-related clinical decisions to minimize such risk
Intelligenza Artificiale e Fattori Umani: esplorazione di nuovi metodi per l’ergonomia fisica in contesti reali
No full textL'ergonomia fisica rappresenta un aspetto cruciale per garantire il benessere e la
sicurezza dei lavoratori, contribuendo al tempo stesso a migliorare la produttività e la
sostenibilità dei sistemi industriali. L'integrazione dei fattori umani nella progettazione
di macchine, sistemi e ambienti di lavoro è fondamentale, soprattutto alla luce dei
paradigmi dell'Industria 4.0 e 5.0, che mettono al centro la cooperazione tra uomo e
macchina. In questo contesto, le tecnologie emergenti di Motion Capture basate su Deep
Learning offrono nuove possibilità per monitorare e migliorare l'ergonomia dei
lavoratori, riducendo il rischio di disturbi muscoloscheletrici legati al lavoro. Questa
ricerca si concentra sull'analisi e il confronto di due approcci distinti per l'utilizzo delle
tecnologie di Deep Learning nel monitoraggio dell'ergonomia fisica dei lavoratori: un
approccio analitico, che utilizza modelli matematici per calcolare il punteggio
ergonomico a partire dagli angoli articolari, e un approccio induttivo, che sfrutta la
capacità predittiva dei sistemi di Machine Learning per determinare il rischio
ergonomico a partire dai dati osservati. L'approccio analitico è stato implementato e
testato attraverso uno strumento chiamato RGB-MAS, che utilizza telecamere RGB per
monitorare i movimenti dei lavoratori e calcolare gli angoli articolari necessari per
l'analisi ergonomica. I risultati dei test, condotti sia in laboratorio che in un contesto
produttivo reale, dimostrano l'efficacia dello strumento nel fornire una valutazione
accurata del rischio ergonomico, nonostante alcune limitazioni legate alla gestione delle
occlusioni e alla precisione dei dati raccolti. L'approccio induttivo è stato invece
implementato attraverso il sistema iRGB-MAS, che utilizza dataset sintetici per
addestrare modelli di Machine Learning capaci di predire il rischio ergonomico sulla
base delle posture rilevate dalle telecamere. Anche in questo caso, i test condotti hanno
evidenziato una buona accuratezza del sistema, suggerendo la possibilità di utilizzare
l'approccio induttivo per migliorare ulteriormente l'efficacia delle valutazioni
ergonomiche, in particolare in contesti dinamici e complessi. I risultati di questa ricerca
contribuiscono a delineare le potenzialità e i limiti delle tecnologie di Motion Capture
basate su Deep Learning per la valutazione dell'ergonomia fisica, offrendo nuove
prospettive per l'implementazione di soluzioni sostenibili e centrate sull'uomo negli
ambienti di lavoro manifatturieri industriali.Physical ergonomics is a crucial aspect of ensuring the well-being and safety of workers,
while contributing to improving the productivity and sustainability of industrial systems. The
integration of human factors in the design of machines, systems and work environments is
crucial, especially in the light of Industry 4.0 and 5.0 paradigms, which focus on cooperation
between man and machine. In this context, emerging Motion Capture technologies based on
Deep Learning offer new possibilities to monitor and improve workers' ergonomics, reducing
the risk of work-related musculoskeletal disorders. This research focuses on the analysis and
comparison of two distinct approaches for the use of Deep Learning technologies in monitoring
the physical ergonomics of workers: an analytical approach, which uses mathematical models
to calculate the ergonomic score from joint angles, and an inductive approach, which exploits
the predictive capacity of Machine Learning systems to determine ergonomic risk from
observed data. The analytical approach was implemented and tested through a tool called RGB-
MAS, which uses RGB cameras to monitor workers' movements and calculate the joint angles
required for ergonomic analysis. The results of the tests, conducted both in the laboratory and
in a real production context, demonstrate the effectiveness of the tool in providing an accurate
assessment of ergonomic risk, despite some limitations related to the handling of occlusions and
the accuracy of the data collected. The inductive approach was instead implemented through
the iRGB-MAS system, which uses synthetic datasets to train Machine Learning models capable
of predicting ergonomic risk based on the postures detected by the cameras. Again, the tests
conducted revealed a good accuracy of the system, suggesting the possibility of using the
inductive approach to further improve the effectiveness of ergonomic assessments, particularly
in dynamic and complex contexts. The results of this research contribute to outlining the
potential and limitations of Motion Capture technologies based on Deep Learning for the
assessment of physical ergonomics, offering new perspectives for the implementation of
sustainable, human-centred solutions in industrial manufacturing workplaces
A new path to code-based signatures via identification schemes with restricted errors
No full textIn this paper, we introduce a new variant of the syndrome decoding problem (SDP), called restricted SDP (R-SDP), in which the entries of the solution vector live in a fixed subset of the underlying finite field. We prove the NP-completeness of R-SDP via a reduction from SDP and show how this new problem can be employed in identification schemes. We revisit some concepts of classical coding theory in light of this new setting, provide several bounds such as the Gilbert-Varshamov, the Singleton, and the Plotkin bound, and study the behavior of random codes. Resulting from this initial work, several proposals have arisen; in particular, the code-based digital signature scheme named CROSS, which is currently competing within NIST's additional call for post-quantum digital signatures, is based on R-SDP
Human head and neck cancer cell lines response to cold atmospheric plasma activated media is affected by the chemistry of culture media
Get PDFSurvival rate of head and neck squamous cell carcinomas (HNSCC) patients are still to date very poor, and the application of innovative clinical approaches are urgently needed. Cold atmospheric plasmas (CAPs) are partially ionized gases that have shown anti-tumor effectiveness over a wide range of cancer types with potential application into clinics. However, the comprehension of the mechanisms underlying indirect CAP effects plays a key role for the prediction of treatment outcomes. In our work, we assessed the potential application of indirect CAP, by using plasma activated media (PAM) and plasma-treated liquids (PTL), as therapeutic strategies for HNSCC treatment. The impact of PAM obtained from exposure to CAP for different times was evaluated in three head and neck cancer (HNC) cell lines (HSC3, FaDu, CAL-27). Cytotoxic effects as inhibition of proliferation, apoptosis rate and cell cycle modifications were tested for the different PAM, showing a time- and cell culture media-dependence tightly related to the chemical composition of PAM considered. In addition, cytotoxic effects were also observed on HNC, two bladder cancer models and one breast cancer cell line when considering PTL, paving the way for their application into a clinical setting
Critical Non-linearity for some Evolution Equations with Fujita-type Critical Exponent
Get PDFWe consider the Cauchy problem for a class of non-linear evolution equations in the form (Formula presented.) here, L(∂t,∂x) is a linear partial differential operator with constant coefficients, of order m≥1 with respect to the time variable t, and l is a natural number satisfying 0≤l≤m-1. For several different choices of L, many authors have investigated the existence of global (in time) solutions to this problem when F(s)=|s|p is a power non-linearity, looking for a critical exponentpc>1 such that global small data solutions exist in the supercritical case p>pc, whereas no global weak solutions exist, under suitable sign assumptions on the data, in the subcritical case
Molecularly imprinted polymers (MIPs) for SARS-CoV-2 omicron variant inhibition: An alternative approach to address the challenge of emerging zoonoses
Get PDFEmerging zoonoses pose significant public health risks and necessitate rapid and effective treatment responses. This study enhances the technology for preparing Molecularly Imprinted Polymers (MIPs), which function as synthetic nanoparticles targeting SARS-CoV-2 receptor-binding domain (RBD), specifically the Omicron variant, thereby inhibiting its function. This study builds on previous findings by introducing precise adjustments in the formulation and process conditions to enhance particle stability and ensure better control over size and distribution, thereby overcoming the issues identified in earlier research. Following docking studies, imprinted nanoparticles were synthesized via inverse microemulsion polymerization and characterized in terms of size, morphology and surface charge. The selective recognition properties and ability of MIPs to obstruct the interaction between ACE2 and the RBD of SARS-CoV-2 were assessed in vitro, using Non-Imprinted Polymers (NIPs) as controls, and rebinding studies were conducted utilizing a Quartz Crystal Microbalance with Dissipation monitoring (QCM-D). The synthesized nanoparticles exhibited uniform dispersion and had a consistent diameter within the nanoscale range. MIPs demonstrated significant recognition properties and exhibited a concentrationdependent ability to reduce RBD binding to ACE2 without cytotoxic or sensitizing effects. MIPs-based platforms offer a promising alternative to natural antibodies for treating SARS-CoV-2 infections, therefore representing a versatile platform for managing emerging zoonoses
Evaluating Sustainability Reporting: A Methodological Approach to Assessing Corporate Disclosure on Diversity & Inclusion
No full textScopo: Questo studio affronta la crescente importanza della CSR (Corporate Social Responsibility) e la necessità di metodi sistematici per valutare la qualità di tale disclosure, in particolare nell’ambito della Diversità e Inclusione (D&I). L’obiettivo della ricerca mira al miglioramento della trasparenza, la fiducia e l’accountability nel reporting di sostenibilità, introducendo un quadro di valutazione completo.
Design/Metodologia/Approccio: Utilizzando la Qualitative Content Analysis, lo studio esamina la disclosure relativa alla D&I su un campione globale di aziende. È stato sviluppato un punteggio di disclosure D&I basato su criteri di valutazione derivati da standard consolidati di sostenibilità (ad esempio, GRI, CSRD). Questa metodologia valuta non solo la quantità, ma anche la profondità, il contenuto e la qualità complessiva delle informazioni divulgate.
Risultati: La ricerca evidenzia significative variazioni nella qualità delle disclosure sulla D&I tra le aziende. Sono state identificate best practice, trend emergenti e aree di miglioramento, dimostrando come disclosure di alta qualità favoriscano la fiducia degli stakeholder e riducano i rischi di "woke washing" e "diversity washing."
Limitazioni della Ricerca: I risultati dello studio sono limitati da un campione ristretto e da occasionali mancanze nei dati disponibili pubblicamente. Le ricerche future potrebbero includere dataset più ampi, come siti web e comunicati stampa aziendali, per rafforzare l’analisi.
Implicazioni Pratiche: Questo studio propone un approccio metodologico che consente alle aziende di valutare in modo sistematico il loro impegno nella D&I e la qualità delle disclosure. Fornisce, inoltre, uno strumento per allineare le disclosure alle aspettative degli stakeholder e migliorare la trasparenza delle performance. Le evidenze empiriche emerse contribuiscono ad avanzare le pratiche di governance aziendale e ad affrontare le discrepanze nelle pratiche di reporting sulla D&I.Purpose: This study addresses the growing emphasis on Corporate Social Responsibility (CSR) disclosures and the need for systematic methods to evaluate the quality of these disclosures, particularly in Diversity and Inclusion (D&I). It aims to enhance transparency, trust, and accountability in corporate sustainability reporting by introducing a comprehensive assessment framework.
Design/Methodology/Approach: The study uses qualitative content analysis to examine D&I disclosures from a global sample of companies. A D&I disclosure grade was developed, incorporating scoring criteria informed by established sustainability standards (e.g., GRI, CSRD). This methodology evaluates the quantity, depth, content, and overall quality of disclosures.
Findings: The research highlights significant variations in the quality of D&I disclosures across companies. It identifies best practices, emerging trends, and areas requiring improvement while demonstrating that high-quality disclosures foster stakeholder confidence and mitigate risks such as “woke washing” and “diversity washing.”
Research Limitations: The study’s findings are limited by a small sample size and occasional gaps in publicly available data. Future research could include broader datasets, such as websites and press releases, to strengthen the analysis.
Practical Implications: This study offers a methodological approach for companies to assess their D&I commitment and disclosure quality systematically. It provides a tool to align disclosures with stakeholder expectations and improve performance transparency. The insights contribute to advancing corporate governance and addressing discrepancies in D&I reporting practices
Investigating female entrepreneurship: a micro-perspective of drivers and barriers for aspiring and experienced women entrepreneurs
No full textGender equality represents a priority goal of the 2030 Agenda for Sustainable Development. Women everywhere run significant businesses, generating positive impacts on society. However, they still face numerous obstacles to starting and sustaining entrepreneurial activities. Extant literature, while increasingly acknowledging the importance of women’s entrepreneurship, often overlooks the nuanced differences between aspirant and experienced female entrepreneurs. This study aims to address this gap by adopting a micro perspective and examining the commonalities and differences in the drivers and barriers to entrepreneurship among aspirant and experienced women entrepreneurs in Italy. Through semi-structured interviews conducted between September and December 2022, this exploratory study sheds light on the factors influencing women’s entrepreneurial endeavors in a developed economy. We identified seven main themes: knowledge as gap for both, the role of technology, external support, socio-cultural background, network ties, territory concern, and personal dimension. Findings underscore the importance of understanding diverse backgrounds, motivations, and challenges faced by women entrepreneurs, offering insights for academic research. While at a general level factors and barriers are similar, at a micro level there are differences that could jeopardize both the success of this path in the start-up phase and its survival in the long term. The study also offers valuable practical implications by highlighting that women entrepreneurs should focus on idea validation, financial resources, networking, knowledge, and cultural stereotypes. At a policy level, tailored support strategies are recommended, including streamlined funding application processes and targeted training initiatives as well as measures to reduce socio-cultural barriers
Application of new rapid methodologies in sensory and consumer science to accelerate the strawberry breeding
No full textLa produzione globale di fragole ha raggiunto 9,36 milioni di tonnellate nel 2021, con l'Asia come principale produttore (4,48 milioni di tonnellate), seguita dalle Americhe (2,22 milioni di tonnellate), dall'Europa (1,76 milioni di tonnellate), dall'Africa (827,21 mila tonnellate) e dall'Oceania (59,72 mila tonnellate). Negli ultimi anni, le preferenze dei consumatori sono cambiate, passando dalla sola quantità alla priorità per la qualità sensoriale e nutrizionale. I principali fattori di attrazione per i consumatori ora includono colore, aroma e gusto. In questo contesto, l'analisi sensoriale emerge come uno strumento fondamentale per comprendere le esigenze dei consumatori, particolarmente durante le prime fasi di selezione dei materiali avanzati di breeding delle fragole.
Questo studio si concentra sullo sviluppo di un lessico sensoriale multilingue per le fragole per future ricerche e sull'implementazione di metodologie sensoriali rapide come Check-All-That-Apply (CATA) e Rate-All-That-Apply (RATA) come alternative efficienti ai metodi sensoriali tradizionali basati su panel addestrati. L'analisi CATA nel breeding delle fragole ha identificato i cultivar più allineati con il prodotto ideale per il consumatore e, attraverso l'analisi delle penalità, ha evidenziato gli attributi visivi e gustativi critici che impattano negativamente sull'accettabilità complessiva e, d'altra parte, quelli che hanno un impatto positivo. I risultati della RATA hanno rivelato distinti driver di gradimento per gli attributi di gusto e visivi: il sapore di fragola, la dolcezza, la succosità e l'aroma erano i principali determinanti dell'accettabilità, mentre la luminosità, la forma regolare e l'omogeneità erano i principali tratti visivi.
Questi risultati forniscono un lessico sensoriale multilingue da utilizzare in futuri studi sensoriali sulle fragole. Inoltre, questo studio sottolinea l'importanza della valutazione sensoriale nei programmi di breeding e convalida l'uso delle metodologie CATA e RATA come strumenti efficaci per ottimizzare il processo di selezione nelle fasi avanzate del breeding delle fragole, garantendo risultati robusti e affidabili.Global strawberry production reached 9.36 million tonnes in 2021, with Asia as the leading producer (4.48 million tonnes), followed by the Americas (2.22 million tonnes), Europe (1.76 million tonnes), Africa (827.21
thousand tonnes), and Oceania (59.72 thousand tonnes). In recent years, consumer preferences have shifted
from focusing solely on quantity to prioritizing sensory and nutritional quality. Key drivers of consumer
attraction now include colour, aroma, and taste. In this context, sensory analysis emerges as a vital tool for
understanding consumer needs, particularly during the early stages of selecting advanced strawberry
breeding materials.
This study focuses on developing a multilingual strawberry sensory lexicon for future research and
implementing rapid sensory methodologies Check-All-That-Apply (CATA) and Rate-All-That-Apply (RATA) as
efficient alternatives to traditional sensory methods based on trained panels.
CATA analysis in strawberry breeding identified cultivars most aligned with the ideal consumer product and,
through penalty analysis, highlighted critical visual and taste attributes that negatively impact overall
acceptability and on the other hand those giving a positive impact. RATA findings revealed distinct drivers of
liking for taste and visual attributes: strawberry flavour, sweetness, juiciness, and aroma were the main
determinants of acceptability, while brightness, regular shape, and homogeneity were key visual traits.
These results provide a multilingual strawberry lexicon to be used in future strawberry sensory studies.
Furthermore, this study underlines the importance of sensory evaluation in breeding programs and validates
the use of CATA and RATA methodologies as effective tools for optimizing the selection process in the
advanced stages of strawberry breeding, ensuring robust and reliable outcomes