arXiv subjects

Paola Velardi

Publications and source records attributed to Paola Velardi.

15 recordsLinked to original sources

arXiv2026

A Multi-Agent Framework for Interpreting Multivariate Physiological Time Series

Continuous physiological monitoring is central to emergency care, yet deploying trustworthy AI is challenging. While LLMs can translate complex physiological signals into clinical narratives, it is unclear how agentic systems perform relative to zero-shot inference. To address these questions, we present Vivaldi, a role-structured multi-agent system that explains multivariate physiological time series. Due to regulatory constraints that preclude live deployment, we instantiate Vivaldi in a controlled, clinical pilot to a small, highly qualified cohort of emergency medicine experts, whose evaluations reveal a context-dependent picture that contrasts with prevailing assumptions that agentic reasoning uniformly improves performance. Our experiments show that agentic pipelines substantially benefit non-thinking and medically fine-tuned models, improving expert-rated explanation justification and relevance by +6.9 and +9.7 points, respectively. Contrarily, for thinking models, agentic orchestration often degrades explanation quality, including a 14-point drop in relevance, while improving diagnostic precision (ESI F1 +3.6). We also find that explicit tool-based computation is decisive for codifiable clinical metrics, whereas subjective targets, such as pain scores and length of stay, show limited or inconsistent changes. Expert evaluation further indicates that gains in clinical utility depend on visualization conventions, with medically specialized models achieving the most favorable trade-offs between utility and clarity. Together, these findings show that the value of agentic AI lies in the selective externalization of computation and structure rather than in maximal reasoning complexity, and highlight concrete design trade-offs and learned lessons, broadly applicable to explainable AI in safety-critical healthcare settings.

cs.LG↗

arXiv2026

Beyond Edge Deletion: A Comprehensive Approach to Counterfactual Explanation in Graph Neural Networks

Graph Neural Networks (GNNs) are increasingly adopted across domains such as molecular biology and social network analysis, yet their black-box nature hinders interpretability and trust. This is especially problematic in high-stakes applications, such as predicting molecule toxicity, drug discovery, or guiding financial fraud detections, where transparent explanations are essential. Counterfactual explanations - minimal changes that flip a model's prediction - offer a transparent lens into GNNs' behavior. In this work, we introduce XPlore, a novel technique that significantly broadens the counterfactual search space. It consists of gradient-guided perturbations to adjacency and node feature matrices. Unlike most prior methods, which focus solely on edge deletions, our approach belongs to the growing class of techniques that optimize edge insertions and node-feature perturbations, here jointly performed under a unified gradient-based framework, enabling a richer and more nuanced exploration of counterfactuals. To quantify both structural and semantic fidelity, we introduce a cosine similarity metric for learned graph embeddings that addresses a key limitation of traditional distance-based metrics, and demonstrate that XPlore produces more coherent and minimal counterfactuals. Empirical results on 13 real-world and 5 synthetic benchmarks show up to +56.3% improvement in validity and +52.8% in fidelity over state-of-the-art baselines, while retaining competitive runtime.

cs.LG↗

arXiv2025

TRADES: Generating Realistic Market Simulations with Diffusion Models

Financial markets are complex systems characterized by high statistical noise, nonlinearity, volatility, and constant evolution. Thus, modeling them is extremely hard. Here, we address the task of generating realistic and responsive Limit Order Book (LOB) market simulations, which are fundamental for calibrating and testing trading strategies, performing market impact experiments, and generating synthetic market data. We propose a novel TRAnsformer-based Denoising Diffusion Probabilistic Engine for LOB Simulations (TRADES). TRADES generates realistic order flows as time series conditioned on the state of the market, leveraging a transformer-based architecture that captures the temporal and spatial characteristics of high-frequency market data. There is a notable absence of quantitative metrics for evaluating generative market simulation models in the literature. To tackle this problem, we adapt the predictive score, a metric measured as an MAE, to market data by training a stock price predictive model on synthetic data and testing it on real data. We compare TRADES with previous works on two stocks, reporting a 3.27 and 3.48 improvement over SoTA according to the predictive score, demonstrating that we generate useful synthetic market data for financial downstream tasks. Furthermore, we assess TRADES's market simulation realism and responsiveness, showing that it effectively learns the conditional data distribution and successfully reacts to an experimental agent, giving sprout to possible calibrations and evaluations of trading strategies and market impact experiments. To perform the experiments, we developed DeepMarket, the first open-source Python framework for LOB market simulation with deep learning. In our repository, we include a synthetic LOB dataset composed of TRADES's generated simulations.

q-fin.TR↗

arXiv2025

AI on the Pulse: Real-Time Health Anomaly Detection with Wearable and Ambient Intelligence

We introduce AI on the Pulse, a real-world-ready anomaly detection system that continuously monitors patients using a fusion of wearable sensors, ambient intelligence, and advanced AI models. Powered by UniTS, a state-of-the-art (SoTA) universal time-series model, our framework autonomously learns each patient's unique physiological and behavioral patterns, detecting subtle deviations that signal potential health risks. Unlike classification methods that require impractical, continuous labeling in real-world scenarios, our approach uses anomaly detection to provide real-time, personalized alerts for reactive home-care interventions. Our approach outperforms 12 SoTA anomaly detection methods, demonstrating robustness across both high-fidelity medical devices (ECG) and consumer wearables, with a ~ 22% improvement in F1 score. However, the true impact of AI on the Pulse lies in @HOME, where it has been successfully deployed for continuous, real-world patient monitoring. By operating with non-invasive, lightweight devices like smartwatches, our system proves that high-quality health monitoring is possible without clinical-grade equipment. Beyond detection, we enhance interpretability by integrating LLMs, translating anomaly scores into clinically meaningful insights for healthcare professionals.

cs.LG↗

arXiv2025

Seamless Monitoring of Stress Levels Leveraging a Universal Model for Time Sequences

Monitoring the stress level in patients with neurodegenerative diseases can help manage symptoms, improve patient's quality of life, and provide insight into disease progression. In the literature, ECG, actigraphy, speech, voice, and facial analysis have proven effective at detecting patients' emotions. On the other hand, these tools are invasive and do not integrate smoothly into the patient's daily life. HRV has also been proven to effectively indicate stress conditions, especially in combination with other signals. However, when HRV is derived from less invasive devices than the ECG, like wristbands and smartwatches, the quality of measurements significantly degrades. This paper presents a methodology for stress detection from a wristband based on a universal model for time series, UniTS, which we finetuned for the task and equipped with explainability features. We cast the problem as anomaly detection rather than classification to favor model adaptation to individual patients and allow the clinician to maintain greater control over the system's predictions. We demonstrate that our proposed model considerably surpasses 12 top-performing methods on three benchmark datasets. Furthermore, unlike other state-of-the-art systems, UniTS enables seamless monitoring, as it shows comparable performance when using signals from invasive or lightweight devices.

cs.LG↗

arXiv2024

Programming Skills are Not Enough: a Greedy Strategy to Attract More Girls to Study Computer Science

It has been observed in many studies that female students in general are unwilling to undertake a course of study in ICT. Recent literature has also pointed out that undermining the prejudices of girls with respect to these disciplines is very difficult in adolescence, suggesting that, to be effective, awareness programs on computer disciplines should be offered in pre-school or lower school age. On the other hand, even assuming that large-scale computer literacy programs can be immediately activated in lower schools and kindergartens, we can't wait for >15-20 years before we can appreciate the effectiveness of these programs. The scarcity of women in ICT has a tangible negative impact on countries' technological innovation, which requires immediate action. In this paper, we describe a strategy, and the details of a number of programs coordinated by the Engineering and Computer Science Departments at Sapienza University, to make high school girl students aware of the importance of new technologies and ICT. In addition to describing the theoretical approach, the paper offers some project examples.

cs.CY↗

arXiv2024

V-RECS, a Low-Cost LLM4VIS Recommender with Explanations, Captioning and Suggestions

NL2VIS (natural language to visualization) is a promising and recent research area that involves interpreting natural language queries and translating them into visualizations that accurately represent the underlying data. As we navigate the era of big data, NL2VIS holds considerable application potential since it greatly facilitates data exploration by non-expert users. Following the increasingly widespread usage of generative AI in NL2VIS applications, in this paper we present V-RECS, the first LLM-based Visual Recommender augmented with explanations(E), captioning(C), and suggestions(S) for further data exploration. V-RECS' visualization narratives facilitate both response verification and data exploration by non-expert users. Furthermore, our proposed solution mitigates computational, controllability, and cost issues associated with using powerful LLMs by leveraging a methodology to effectively fine-tune small models. To generate insightful visualization narratives, we use Chain-of-Thoughts (CoT), a prompt engineering technique to help LLM identify and generate the logical steps to produce a correct answer. Since CoT is reported to perform poorly with small LLMs, we adopted a strategy in which a large LLM (GPT-4), acting as a Teacher, generates CoT-based instructions to fine-tune a small model, Llama-2-7B, which plays the role of a Student. Extensive experiments-based on a framework for the quantitative evaluation of AI-based visualizations and on manual assessment by a group of participants-show that V-RECS achieves performance scores comparable to GPT-4, at a much lower cost. The efficacy of the V-RECS teacher-student paradigm is also demonstrated by the fact that the un-tuned Llama fails to perform the task in the vast majority of test cases. We release V-RECS for the visualization community to assist visualization designers throughout the entire visualization generation process.

cs.HC↗

arXiv2024

Agnostic Visual Recommendation Systems: Open Challenges and Future Directions

Visualization Recommendation Systems (VRSs) are a novel and challenging field of study aiming to help generate insightful visualizations from data and support non-expert users in information discovery. Among the many contributions proposed in this area, some systems embrace the ambitious objective of imitating human analysts to identify relevant relationships in data and make appropriate design choices to represent these relationships with insightful charts. We denote these systems as "agnostic" VRSs since they do not rely on human-provided constraints and rules but try to learn the task autonomously. Despite the high application potential of agnostic VRSs, their progress is hindered by several obstacles, including the absence of standardized datasets to train recommendation algorithms, the difficulty of learning design rules, and defining quantitative criteria for evaluating the perceptual effectiveness of generated plots. This paper summarizes the literature on agnostic VRSs and outlines promising future research directions.

cs.LG↗

arXiv2023

Unsupervised Detection of Behavioural Drifts with Dynamic Clustering and Trajectory Analysis

Real-time monitoring of human behaviours, especially in e-Health applications, has been an active area of research in the past decades. On top of IoT-based sensing environments, anomaly detection algorithms have been proposed for the early detection of abnormalities. Gradual change procedures, commonly referred to as drift anomalies, have received much less attention in the literature because they represent a much more challenging scenario than sudden temporary changes (point anomalies). In this paper, we propose, for the first time, a fully unsupervised real-time drift detection algorithm named DynAmo, which can identify drift periods as they are happening. DynAmo comprises a dynamic clustering component to capture the overall trends of monitored behaviours and a trajectory generation component, which extracts features from the densest cluster centroids. Finally, we apply an ensemble of divergence tests on sliding reference and detection windows to detect drift periods in the behavioural sequence.

cs.LG↗

arXiv2023

LOB-Based Deep Learning Models for Stock Price Trend Prediction: A Benchmark Study

The recent advancements in Deep Learning (DL) research have notably influenced the finance sector. We examine the robustness and generalizability of fifteen state-of-the-art DL models focusing on Stock Price Trend Prediction (SPTP) based on Limit Order Book (LOB) data. To carry out this study, we developed LOBCAST, an open-source framework that incorporates data preprocessing, DL model training, evaluation and profit analysis. Our extensive experiments reveal that all models exhibit a significant performance drop when exposed to new data, thereby raising questions about their real-world market applicability. Our work serves as a benchmark, illuminating the potential and the limitations of current approaches and providing insight for innovative solutions.

q-fin.TR↗

arXiv2022

AI-based Data Preparation and Data Analytics in Healthcare: The Case of Diabetes

The Associazione Medici Diabetologi (AMD) collects and manages one of the largest worldwide-available collections of diabetic patient records, also known as the AMD database. This paper presents the initial results of an ongoing project whose focus is the application of Artificial Intelligence and Machine Learning techniques for conceptualizing, cleaning, and analyzing such an important and valuable dataset, with the goal of providing predictive insights to better support diabetologists in their diagnostic and therapeutic choices.

cs.LG↗

arXiv2021

A network-based analysis of disease modules from a taxonomic perspective

Objective: Human-curated disease ontologies are widely used for diagnostic evaluation, treatment and data comparisons over time, and clinical decision support. The classification principles underlying these ontologies are guided by the analysis of observable pathological similarities between disorders, often based on anatomical or histological principles. Although, thanks to recent advances in molecular biology, disease ontologies are slowly changing to integrate the etiological and genetic origins of diseases, nosology still reflects this "reductionist" perspective. Proximity relationships of disease modules (hereafter DMs) in the human interactome network are now increasingly used in diagnostics, to identify pathobiologically similar diseases and to support drug repurposing and discovery. On the other hand, similarity relations induced from structural proximity of DMs also have several limitations, such as incomplete knowledge of disease-gene relationships and reliability of clinical trials to assess their validity. The purpose of the study described in this paper is to shed more light on disease similarities by analyzing the relationship between categorical proximity of diseases in human-curated ontologies and structural proximity of the related DM in the interactome. Method: We propose a methodology (and related algorithms) to automatically induce a hierarchical structure from proximity relations between DMs, and to compare this structure with a human-curated disease taxonomy. Results: We demonstrate that the proposed methodology allows to systematically analyze commonalities and differences among structural and categorical similarity of human diseases, help refine and extend human disease classification systems, and identify promising network areas where new disease-gene interactions can be discovered.

q-bio.MN↗

arXiv2019

Network-based methods for disease-gene prediction

We predict disease-genes relations on the Human Interactome network using a methodology that jointly learns functional and connectivity patterns surrounding proteins. Contrary to other data structures, the Interactome is characterized by high incompleteness and absence of explicit negative knowledge, which makes predictive tasks particularly challenging. To exploit at best latent information in the network, we propose an extended version of random walks, named Random Watcher-Walker ($RW^2$), which is able to learn rich representations of disease genes (or gene products) features. Our method successfully compares with the best known system for disease gene prediction, and other state-of-the-art graph-based methods. We perform sensitivity analysis and apply perturbations to ensure robustness. In contrast with previous studies, our results demonstrate that connectivity alone is not sufficient to classify disease-related genes.

q-bio.MN↗

arXiv2019

Quality of Life Assessment of Diabetic patients from health-related blogs

Motivations: People are generating an enormous amount of social data to describe their health care experiences, and continuously search information about diseases, symptoms, diagnoses, doctors, treatment options and medicines. The increasing availability of these social traces presents an interesting opportunity to enhance timeliness and efficiency of care. By collecting, analyzing and exploiting this information, it is possible to modify or in any case significantly improve our knowledge on the manifestation of a pathology and obtain a more detailed and nuanced vision of patients' experience, that we call the "social phenotype" of diseases. Materials and methods: In this paper we present a data analytic framework to represent, extract and analyze the social phenotype of diseases. To show the effectiveness of our methodology we presents a detailed case study on diabetes. First, we create a high quality data sample of diabetic patients' messages, extracted from popular medical forums during more than 10 years. Next, we use a topic extraction techniques based on latent analysis and word embeddings, to identify the main complications, the frequently reported symptoms and the common concerns of these patients. Results: We show that a freely manifested perception of a disease can be noticeably different from what is inferred from questionnaires, surveys and other common methodologies used to measure the impact of a disease on the patients' quality of life. In our case study on diabetes, we found that issues reported to have a daily impact on diabetic patients are diet, glycemic control, drugs and clinical tests. These problems are not commonly considered in Quality of Life assessments, since they are not perceived by doctors as representing severe limitations.

cs.CY↗

arXiv2015

Analysis of women leadership in enterprise social networks

This paper describes a Social Network Analysis toolkit to monitor an Enterprise Social Network and help analyzing informal leadership as a function of social ties and topic discussions. The toolkit has been developed in the context of a regional project, Fiordaliso, funded by Regione Lazio (a region of central Italy) and leaded by Reply, an international network of specialized companies in the field of digital services.

cs.SI↗