Lidia Marassi

Lidia Marassi

Auto-generated profile, based on publicly available information

Lidia Marassi


Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
20 May 2025 in AI
Verified
Journal Article
Open access
PDF file
Ojasvi GuptaOjasvi Gupta
Ojasvi Gupta
SMStefano Marrone
Stefano Marrone
Francesco GargiuloFrancesco Gargiulo
Francesco Gargiulo
RJRajesh Jaiswal
Rajesh Jaiswal
LMLidia Marassi
Lidia Marassi
Background/Objectives: Large Language Models (LLMs) like ChatGPT, LLAMA, and Mistral are widely used for automating tasks such as content creation and data analysis. However, due to their training on publicly available internet data, they may inherit social biases. We aimed to investigate the social biases (i.e., ethnic, gender, and disability biases) in these models and evaluate how different model versions handle them. Methods: We instruction-tuned popular models (like Mistral, LLAMA, and Gemma), and for this we curated a dataset constructed by collecting and modifying diverse data from various public datasets. Prompts were run through a controlled pipeline, and responses were categorized (e.g., biased, confused, repeated, or accurate) and analyzed. Results: We found that models responded differently to bias prompts depending on their version. Fine-tuned models showed fewer overt biases but more confusion or censorship. Disability-related prompts triggered the most consistent biases across models. Conclusions: Bias persists in LLMs despite instruction tuning. Differences between model versions may lead to inconsistent user experiences and hidden harms in downstream applications. Greater transparency and robust fairness testing are essential.

Verified
Journal Article
Open access
PDF file
AAanna annunziata
anna annunziata
SCSalvatore Cappabianca
Salvatore Cappabianca
Salvatore CapuozzoSalvatore Capuozzo
Salvatore Capuozzo
Nicola CoppolaNicola Coppola
Nicola Coppola
CSCamilla Di Somma
Camilla Di Somma
Carlo SansoneCarlo Sansone
Carlo Sansone
Hospital overcrowding, driven by both structural management challenges and widespread medical emergencies, has prompted extensive research into machine learning (ML) solutions for predicting patient length of stay (LOS) to optimize bed allocation. While many existing models simplify the LOS prediction problem to a classification task, predicting broad ranges of hospital days, an exact day-based regression model is often crucial for precise planning. Additionally, available data are typically limited and heterogeneous, often collected from a small patient cohort. To address these challenges, we present a novel multimodal ML framework that combines imaging and clinical data to enhance LOS prediction accuracy. Specifically, our approach uses the following: (i) feature extraction from chest CT scans via a convolutional neural network (CNN), (ii) their integration with clinically relevant tabular data from patient exams, refined through a feature selection system to retain only significant predictors. As a case study, we applied this framework to pneumonia patient data collected during the COVID-19 pandemic at two hospitals in Naples, Italy—one specializing in infectious diseases and the other general-purpose. Under our experimental setup, the proposed system achieved an average prediction error of only three days, demonstrating its potential to improve patient flow management in critical care environments.

14 December 2023
Verified
Conference Paper
Open access
LMLidia Marassi
Lidia Marassi
Recently, generative Artificial Intelligence (AI) models have experienced an incredible surge in interest and use. The proliferation of these technologies suggests that it would be prudent to raise awareness of the potential consequences of irresponsible use of these models. Indeed, to achieve trustworthy AI solutions, it is essential to promote AI education as a fundamental step. Educating people to use AI responsibly not only improves their understanding of the technology, but also equips them to address issues of bias and discrimination. Indeed, informed users are more likely to recognize when AI is producing biased or discriminatory results and to demand appropriate solutions. This awareness of the ethical implications of AI decisions is crucial for the responsible and socially conscious adoption of AI. The aim of this poster is to report the results obtained as a final thesis work, pursued within the context of the Human-Centred AI Masters (HCAIM) programme in Naples, focused on the analysis of users’ perceptions of biases in media created by generative AI models.


14 December 2023
Verified
Conference Paper
Open access
LMLidia Marassi
Lidia Marassi
APAntonio Elia Pascarella
Antonio Elia Pascarella
GGGiovanni Giacco
Giovanni Giacco
MZMonica Zuccarini
Monica Zuccarini
SMStefano Marrone
Stefano Marrone
MRMattia Rigiroli
Mattia Rigiroli
The proliferation of voluntary carbon markets (VCMs) as a key tool for incentivising action to reduce greenhouse gas emissions has necessitated the creation of reliable and transparent monitoring mechanisms. This poster focuses on the potential of using Earth observation data and artificial intelligence algorithms for monitoring, reporting and verification of carbon projects in the context of VCMs. The poster is divided into a first technical part, which shows the potential of using Earth observation data and AI for carbon sequestration estimation, and a second part, which considers the main ethical implications of these solutions for VCMs in light of the current European regulations.

2 December 2024
Verified
Conference Paper
Open access
Ojasvi GuptaOjasvi Gupta
Ojasvi Gupta
RJRajesh Jaiswal
Rajesh Jaiswal
SMStefano Marrone
Stefano Marrone
LMLidia Marassi
Lidia Marassi
Francesco GargiuloFrancesco Gargiulo
Francesco Gargiulo