322590 research outputs found
Sort by
Sailing toward healing. A combined group intervention to improve well-being in breast cancer survivors: a brief report study
No full textBreast cancer survivors often face persistent physical and emotional challenges. Evidence suggests that physical exercise and group psychological interventions can improve well-being and illness adjustment. This pilot study examined the effectiveness of a combined intervention delivered in a natural environment. Sixty female breast cancer survivors (Mage = 51.0; SD = 5.5) participated in a 1-week program consisting of daily sailing lessons and group psychological sessions designed to address cancer-related issues. Assessments were conducted 1 week before and 1 week after the intervention using the Functional Assessment of Cancer Therapy-Breast (FACT-B) to assess quality of life and the State-Trait Anxiety Inventory. Linear mixed-effects models tested changes over time in quality of life and anxiety and whether previous psychotherapy or physical activity influenced these outcomes. FACT-B total scores significantly improved over time (P = 0.004), with gains in physical well-being (P < 0.001), emotional well-being (P < 0.001), and breast cancer-specific concerns (P = 0.018). No changes over time were observed in social or functional well-being. Anxiety levels remained moderate and stable (P = 0.250). Previous psychotherapy and physical activity did not significantly influence changes in quality of life or anxiety. A short-term group intervention combining sailing lessons and psychological sessions in a natural environment may enhance quality of life in breast cancer survivors, particularly in physical and emotional domains. These findings suggest that integrative approaches addressing both physical and psychological health are feasible and warrant further investigation in larger controlled studies
ADVANCES IN MEAN-FIELD OPTIMAL CONTROL VIA DYNAMIC PROGRAMMING EQUATIONS
No full textIn questa tesi di dottorato ci proponiamo di analizzare varie classi di problemi di controllo ottimo di tipo McKean-Vlasov con un approccio analitico, che consiste nella caratterizzazione della funzione valore di ciascun problema mediante un'opportuna equazione alle derivate parziali, detta equazione di Hamilton-Jacobi-Bellman (HJB). Poiché i coefficienti di un'equazione stocastica di McKean-Vlasov dipendono non solo dalle traiettorie della soluzione, ma anche dalla sua legge, l'equazione di HJB associata è definita sullo spazio di Wasserstein delle misure di probabilità di momento secondo finito P2(Rd). In questa tesi, la buona positura di tale equazione in senso viscoso è sempre analizzata direttamente sullo spazio di Wasserstein; non ricorriamo pertanto al lifting dell'equazione su uno spazio di Hilbert di variabili aleatorie di quadrato sommabili, in contrasto con una parte della letteratura su questo argomento.
Nel primo capitolo della tesi, formuliamo e analizziamo un problema di controllo ottimo di tipo McKean-Vlasov con orizzonte temporale infinito e funzionale di guadagno scontato. Il problema è ben posto sotto ipotesi standard per i coefficienti dell'equazione di stato, la funzione di guadagno e il fattore di sconto β. Innanzitutto, dimostriamo che la funzione valore non dipende dal tempo iniziale della dinamica. Questa proprietà, spesso considerata naturale per problemi con orizzonte temporale infinito e fattore di sconto, non è invece banale in presenza di effetti di campo medio: in particolare, esibiamo un esempio in cui la proprietà non vale a causa della dipendenza non lineare della funzione di guadagno dalla legge del processo di stato. Tuttavia, assumendo che la funzione di guadagno sia sufficientemente regolare, estendiamo questa proprietà di invarianza temporale al nostro problema di tipo McKean-Vlasov. Successivamente, mostriamo che la funzione valore dipende dalla condizione iniziale della dinamica solo mediante la sua legge. Il valore del problema può dunque essere identificato con una funzione v indipendente dal tempo e definita su P2(Rd). Infine, sullo stesso spazio deriviamo un'opportuna equazione ellittica di Hamilton-Jacobi-Bellman e caratterizziamo completamente la funzione valore v come l'unica soluzione viscosa di tale equazione.
Il secondo capitolo presenta un problema di controllo ottimo con la stessa dinamica di tipo McKean-Vlasov su orizzonte temporale infinito, ma un funzionale di guadagno in forma ergodica. Sotto opportune ipotesi di dissipatività, le soluzioni dell'equazione di stato soddisfano condizioni forti di stabilità rispetto alle condizioni iniziali, dette stime ergodiche. Basandoci su una famiglia di problemi di controllo ausiliari con funzionale di guadagno scontato, costruiamo un valore reale λ e una funzione φ: P2(Rd) → R che costituiscono una soluzione viscosa di un'equazione ellittica, detta equazione ergodica di HJB, sullo spazio P2(Rd). Inoltre, dimostriamo che la coppia (λ,φ) caratterizza il comportamento asintotico delle funzioni valore associate a problemi di controllo ottimo di tipo McKean-Vlasov ad orizzonte temporale finito quando quest'ultimo tende all'infinito e delle funzioni valore associate a problemi di controllo ottimo di tipo McKean-Vlasov ad orizzonte temporale infinito con fattore di sconto quando questo tende a 0. Infine, identifichiamo λ con il valore del problema di controllo ergodico, dimostrando che esso è costante. Presentiamo poi un teorema di verifica che permette di costruire controlli ottimi per il problema ergodico a partire da soluzioni dell'equazione di HJB ergodica. A differenza di quanto proposto da altri autori, non deriviamo l'equazione ergodica di HJB usando proprietà di stabilità di soluzioni viscose di PDE ellittiche, ma tutti i nostri risultati principali seguono da una relazione funzionale soddisfatta dalla coppia (λ,φ).
Il terzo capitolo presenta alcuni risultati preliminari riguardanti un problema di controllo ottimo di tipo McKean-Vlasov con osservazione parziale: la dinamica è infatti influenzata da ua catena di Markov M a tempo continuo e spazio degli stati finito che non è osservata dall'agente. L'analisi del problema richiede dunque tecniche di filtraggio, basate su un problema di controllo ausiliario, detto problema separato, con un'equazione di stato controllata a valori nello spazio delle misure positive (l'equazione di Zakai). Tale equazione si può poi riscrivere come una SDE a valori in RN, dal momento che la catena può assumere solo un numero finito N di valori. Adottando questo approccio, a causa della presenza di effetti di campo medio nel problema originale, otteniamo un problema di controllo di tipo McKean-Vlasov non standard, i cui coefficienti dipendono dalla legge congiunta del processo di stato e del moto Browniano. Studiamo dunque la buona positura di tale problema e la stabilità delle soluzioni dell'equazione di stato rispetto alle condizioni iniziali e dimostriamo che esso è equivalente ad una versione debole del problema originale. Presentiamo infine gli sviluppi futuri di questa ricerca, ricordando che intendiamo adottare un approccio di tipo analitico al problema, che porti alla caratterizzazione del valore mediante un'opportuna equazione di HJB su uno spazio di Wasserstein.This Thesis is devoted to the study of several classes of optimal control problems of the McKean-Vlasov type via an analytic approach, which consists in characterizing the value of each problem by means of a suitable partial differential equation, called Hamilton-Jacobi-Bellman (HJB) equation. As the coefficients of a McKean-Vlasov stochastic differential equation depend on the state process also through its law, the associated HJB equation is defined on the Wasserstein space of probability measures with finite second moment P2(Rd). It is our purpose to discuss the well-posedness of this PDE in a suitable viscosity sense directly on the Wasserstein space, without lifting it to a Hilbert space of square-integrable random variables L2, in contrast with previous contributions.
The first chapter of this Thesis focuses on the formulation and the analysis of a general optimal control problem of the McKean-Vlasov type with infinite time horizon and discounted gain functional. Under standard assumptions on the coefficients of the state equation, the gain function and the discount rate β, the optimization problem is well-posed. We firstly prove that the value function does not depend on the initial time of the dynamics. This property, which is often considered natural for discounted control problems over an unbounded time interval, is not trivial when mean-field effects occur: we provide an example where the property fails, due to the non-linear dependence of the gain function on the law of the state process. Nonetheless, assuming that the gain function is sufficiently smooth, we extend this time invariance property to our McKean-Vlasov type problem. We then show that the value function depends on the initial condition of the dynamics only through its law. The value of the problem can thus be identified with a time-independent function v defined on P2(Rd). Hence, a suitable elliptic Hamilton-Jacobi-Bellman equation is derived on the same space; the value v is completely characterized as its unique solution in a suitable viscosity sense.
The second Chapter presents a different optimal control problem with the same McKean-Vlasov dynamics, infinite time horizon but gain functional in ergodic (long time average) form. Under suitable dissipativity assumptions, we derive strong stability properties of the solutions to the state equation, called ergodic estimates. We use the vanishing discount method, relying on auxiliary control problems with discounted gain functional, to construct a real value λ and a function φ: P2(Rd) → R, which provide a viscosity solution to an ergodic HJB equation on the space P2(Rd). Moreover, the pair (λ,φ) is proved to characterize the asymptotic behaviour of the value functions of finite horizontal mean-field control problems when the time horizon tends to infinity and of infinite horizontal mean-field control problems with discounted gain when the discount rate tends to 0. We eventually identify λ with the value function of the ergodic control problem, thus showing that it is constant. We also present a verification theorem to construct optimal controls for the ergodic optimal control problem by means of a solution to the ergodic HJB equation in closed-loop form. In contrast with previous contributions, we do not derive the ergodic HJB equation using stability properties of solutions to elliptic PDEs, but our main results follow from a fundamental functional relation involving φ and λ.
The third chapter presents some preliminary results concerning an optimal control problem of the McKean-Vlasov type with partial observation: the mean-field dynamics is influenced by a continuous-time Markov chain M with finite state space which can not be observed by the representative agent. The analysis of the problem thus requires filtering techniques, relying on a fully observable auxiliary control problem, called separated problem, with a measure-valued controlled state equation (Zakai equation); this SDE can be rewritten as an RN-valued state equation as M can only assume N values. It is then proved that the separated control problem is equivalent to the original control problem with partial observation in a weak form. Adopting this approach, due to the presence of mean-field effects in the original state equation, we obtain a non-standard optimal control problem of the McKean-Vlasov type, whose coefficients depend on the joint law of the state equation and the Brownian noise. We thus study its well-posedness and stability properties. We also discuss future research with the aim of adopting the analytic approach, eventually leading to a suitable HJB equation on a Wasserstein space
Dialogical AI for cognitive bias mitigation in medical diagnosis
No full textLarge Language Models (LLMs) promise to enhance clinical decision-making, yet empirical studies reveal a paradox: physician performance with LLM assistance shows minimal improvement or even deterioration. This failure stems from an “acquiescence problem”: current LLMs passively confirm rather than challenge clinicians’ hypotheses, reinforcing cognitive biases such as anchoring and premature closure. To address these limitations, we propose a Dialogic Reasoning Framework that operationalizes Dialogical AI principles through a prototype implementation named “Diagnostic Dialogue” (DiDi). This framework operationalizes LLMs into three user-controlled roles: the Framework Coach (guiding structured reasoning), the Socratic Guide (asking probing questions), and the Red Team Partner (presenting evidence-based alternatives). Built upon Retrieval-Augmented Generation (RAG) architecture for factual grounding and traceability, this framework transforms LLMs from passive information providers into active reasoning partners that systematically mitigate cognitive bias. We evaluate the feasibility and qualitative impact of this framework through a pilot study (DiDi) deployed at Centro Chirurgico Toscano (CCT). Through purposive sampling of complex clinical scenarios, we present comparative case studies illustrating how the dialogic approach generates necessary cognitive friction to overcome acquiescence observed in standard LLM interactions. While rigorous clinical validation through randomized controlled trials remains necessary, this work establishes a methodological foundation for designing LLM-based clinical decision support systems that genuinely augment human clinical reasoning
Longitudinal observation of severe asthma comorbidities and oral corticosteroids use from SANI and ISAR registries
No full textBackground
The International Severe Asthma Registry (ISAR) reported a high rate of comorbidities differentially associated with clinical characteristics, biomarkers, and outcomes.
Methods
We aimed to compare the prevalences of comorbidities between global (ISARWORLD) and ITALY-derived ISAR cohorts and to explore characteristics of severe asthma (SA) patients progressively enrolled into the Severe Asthma Network Italy (SANI) registry over 5 years.
Results
T2-related SA comorbidities, including allergic rhinitis (AR), chronic rhinosinusitis (CRS) and nasal polyps (NPs) were more frequent (p < 0.001) in the ITALY cohort in addition to some oral corticosteroids (OCS)-related comorbidities, likely relating to the higher burden of OCS use. A comorbidity-dependent pattern of association for biomarkers and clinical outcomes with AR, CRS and NPs was identified in both ITALY-derived and ISARWORLD cohorts. In addition, a progressive decrease in the frequency of atopy, total IgE, number of exacerbations (AEs), chronic OCS treatment (p < 0.001) and a progressive increase in lung function and eosinophils count was reported longitudinally in the SANI registry. When stratifying by the presence of NPs, sex and smoking status, similar enrolment changes were identified with the additional findings of increased FeNO in NPs and Female cohorts and atopic eczema in smokers.
Conclusion
Longitudinal observation of enrolment characteristics from the Italian SANI registry and comparison with ISAR highlight changes influenced not only by regional population traits but also by the attitude of clinicians, biologics availability and eligibility and the OCS stewardship campaign.
Trial registration
The International Severe Asthma Registry (ISAR): EU PAS number EUPAS23651; Study ID 47596; registered April 16, 2018. Severe Asthma Network Italy (SANI): NCT number: NCT06625216. Retrospectively registered 2024-07-05
Karst record of Holocene climate and human-induced changes in surface processes in the northern Apennines of Italy
No full textUnderground karst is a sensitive component of the Earth's Critical Zone (ECZ), capable of preserving changes in its dynamics driven by both natural and anthropogenic forces. In this study, we examine the Holocene sedimentary archive preserved in Tana della Mussina Cave (TdMC) in the northern Apennines of Italy. This archive includes clastic and chemical (speleothem) sediments, as well as archaeological deposits linked to human occupation of the site since the Copper Age. Our reconstruction indicates that sedimentation in the cave responded to changes in the hydrology of the karst system, alternating between phases of clastic deposition and periods of sedimentary hiatus, along with a documented phase of speleothem precipitation. Clastic sedimentation occurred particularly after the Greenlandian/Northgrippian boundary and again at the onset of the Meghalayan phase of the Holocene, likely in response to increased water availability, which led to the accumulation of fluvial sediments transported through the TdMC catchment. In contrast, speleothem deposition took place later, between 3800 and 2200 years BP. From around 5500 years BP, clastic sedimentation is marked by a significant accumulation of charcoal and lumps of heated sediment/soil. We interpret this evidence because of human activity, specifically extensive deforestation, which has been archaeologically documented in the region since the Copper Age. The slash-and-burn technique, used to clear land for pasture or agriculture, increased soil erosion rates, leading to a greater influx of sediments into the karst system, including charcoal and remnants of heated soils. This case study provides an example of early human land use change impacting surface processes and influencing the dynamics of the ECZ
BCLC strategy for prognosis prediction and treatment recommendations: The 2026 update
No full textThe 2026 BCLC update incorporates recent therapeutic advances in HCC into its recommendations, while preserving simplicity. Each disease stage remains directly linked to its evidence-based first-line treatment option. The BCLC algorithm provides guidance for evaluating patient prognosis and proposes a therapeutic strategy according to the current scientific evidence. Tumour boards require structured methodologies to address clinical complexity. Beyond published evidence, decision-making should incorporate biological, psychosocial, and contextual factors that may affect morbidity, feasibility, and patient vulnerability. Such a multidimensional approach ensures treatments remain evidence-based and patient-centred. The updated clinical decision-making chapter embeds the CUSE (Complexity, Uncertainty, Subjectivity, Emotion) framework. CUSE turns unavoidable doubt into a shared, iterative process to: (i) define the therapeutic goal (survival, tumour control, quality of life, etc .); (ii) grade each option, noting evidence strength and gaps; (iii) align choices with comorbidities, feasibility, oncologic risk, and patient values and goals; and (iv) select a plan with regular check-ins as new information or needs arise
Clinical Trial Eligibility in Atopic Dermatitis: Data from a Large Real-World International Cohort
No full textIntroduction: Randomized clinical trials (RCTs) in atopic dermatitis (AD) often exclude older adults and patients with comorbidities, limiting the generalizability of trial findings to real-world populations. Despite the growing use of biologics and Janus kinase inhibitors (JAKis) in clinical practice, the extent to which real-world patients meet RCT eligibility criteria and their associated safety outcomes remains unclear. Methods: We conducted a retrospective analysis of a large, multicenter international cohort of adolescents and adults with AD treated with biologics (dupilumab, tralokinumab) or systemic JAKis (abrocitinib, baricitinib, upadacitinib) between October 2017 and March 2023 across 16 dermatology centers. Eligibility was defined according to commonly applied RCT criteria. Patients meeting ≥ 1 exclusion criterion were classified as ineligible. Demographic, clinical, and safety outcomes were analyzed. Results: Among 2154 patients, 514 (23.9%) were ineligible. The most frequent reasons were Eczema Area and Severity Index (EASI) < 16 (67.9%), age ≥ 75 years (21.8%), and cardiovascular disease (13.0%). Ineligibility patterns differed across treatments: patients receiving JAKis were most often ineligible because of EASI < 16 (92.8%), whereas ineligibility among biologic users more commonly reflected also age or cardiovascular comorbidity. Ineligible patients were older, had more non-atopic comorbidities, and had lower measured baseline disease activity. Safety profiles were generally favorable. Among biologic-treated patients, adverse event rates were similar between eligible and ineligible groups. In the JAKi cohort, overall adverse events were more frequent in ineligible patients (52.9% vs. 42.6%; p = 0.051), with acneiform eruption and lipid abnormalities emerging as the most distinct differences. No unexpected safety signals were identified in either treatment group. Conclusion: Nearly one-quarter of real-world patients with AD would not have met eligibility criteria for pivotal RCTs, yet both biologics and JAK inhibitors demonstrated acceptable safety profiles in these populations. While adverse events were more frequent among ineligible patients receiving JAKis, findings require further investigation to determine their clinical relevance, particularly regarding long-term cardiovascular outcomes
Physical activity in patients with atrial fibrillation and adverse outcomes: results from a large-scale prospective multicenter European cohort
No full textIntroduction and objectives: Increased physical activity may improve outcomes in patients with atrial fibrillation (AF), but available data are limited. This study aimed to analyze the association between physical activity and major adverse outcomes in a large European AF cohort. Methods: We used data from a prospective, observational, large-scale multicenter study of AF in European patients, involving 27 countries and 250 centers. Patients self-reported their physical activity levels during the preceding 2 years, categorized as none (physically inactive), occasional, regular, or intense. The primary outcome was a composite of major adverse cardiovascular (CV) events and all-cause mortality. Results: A total of 9525 patients with AF were included (median age 71 years; 41% female). Overall, 43% reported being physically inactive. Among those who reported physical activity, 56.9% described it as occasional, 36.5% as regular, and 6.6% as intense. Inactive patients had a higher burden of both CV and non-CV comorbidities. Over a median follow-up of 728 days, 16.2% of the patients experienced the primary composite endpoint. Physical activity was independently associated with a lower risk of the primary endpoint (adjusted HR, 0.68; 95% CI, 0.59-0.76), with a progressively lower incidence as activity levels increased. Interaction analyses showed that the beneficial effect of physical activity was consistent among several subgroups, with no significant interactions. Conclusions: Physical activity, even at low levels, was associated with a reduced incidence of CV events and all-cause mortality. These benefits appear to be independent of CV risk factors and comorbidities, highlighting the importance of incorporating physical activity into holistic AF management strategies