De Angelis P, Andalò A, Gentili N
… +7 more, Giorgetti L, Ridolfi L, Pasolini R, Pagliarani A, Cavallucci M, Vespignani R, Carbonaro A
Recenti Prog Med
· 2025 Oct · PMID 41037379
·
Publisher ↗
Cancer Virtual Lab is a secure and interoperable platform for oncology research. It integrates HL7 FHIR and ontology-based knowledge graphs to structure clinical data, enabling advanced semantic reasoning. Generative AI...Cancer Virtual Lab is a secure and interoperable platform for oncology research. It integrates HL7 FHIR and ontology-based knowledge graphs to structure clinical data, enabling advanced semantic reasoning. Generative AI supports researchers by guiding data exploration and interpretation, accelerating insights and enhancing precision oncology in a privacy-preserving, scalable environment.
De Vita V, Destro Castaniti B, Vassalli M
… +8 more, De Mori L, Lacalaprice D, Arcà E, Cristiano A, Battipaglia C, Risuleo PE, Dionisi T, Causio FA
Recenti Prog Med
· 2025 Oct · PMID 41037378
·
Publisher ↗
Large language models (LLMs) show promise in explicit reasoning for complex medical fields like psychiatry. This study assessed the clinical validity of Gemini's chain-of-thought (CoT) reasoning in 10 complex psychiatric...Large language models (LLMs) show promise in explicit reasoning for complex medical fields like psychiatry. This study assessed the clinical validity of Gemini's chain-of-thought (CoT) reasoning in 10 complex psychiatric cases, evaluated by specialists using six metrics. Results indicate high performance (average score ≥4.26/5), especially in step sufficiency and factual accuracy, suggesting that CoT reasoning by LLMs can support transparent and detailed clinical decision-making.
Pinciroli T, De Angelis L, Causio FA
… +7 more, Iacuzio A, Grieco V, Agrimi E, Traglia F, Valetto MR, Dri P, Naldi L
Recenti Prog Med
· 2025 Oct · PMID 41037377
·
Publisher ↗
Using large language models (LLMs) to answer patients' questions is promising but calls for caution and careful validation. We assessed GPT-4o mini-generated answers to dermatology-related patient questions and compared...Using large language models (LLMs) to answer patients' questions is promising but calls for caution and careful validation. We assessed GPT-4o mini-generated answers to dermatology-related patient questions and compared them to specialists' answers. While GPT-4o mini provided clearer and more updated information, some answers contained inaccuracies that could lead to risky patient management. Future work should explore retrieval techniques from reliable sources to enhance safety and accuracy.
Coro F, Ravizza A, Bonatti AF
… +2 more, De Maria C, Vozzi G
Recenti Prog Med
· 2025 Oct · PMID 41037376
·
Publisher ↗
The use of artificial intelligence in medicine requires software user interfaces that ensure usability, transparency, and compliance with the European AI Act. The proposed framework defines minimum technical requirements...The use of artificial intelligence in medicine requires software user interfaces that ensure usability, transparency, and compliance with the European AI Act. The proposed framework defines minimum technical requirements for each usage phase, from data input to model interpretation, promoting safety, explainability, and human oversight, elements still underdeveloped in clinical practice.
Capuzzi S, Baldisseri F, Cacchione A
… +6 more, Carai A, Fabozzi F, Pietrabissa A, Mastronuzzi A, Tozzi AE, Ferro D
Recenti Prog Med
· 2025 Oct · PMID 41037375
·
Publisher ↗
This study presents a two-phase AI-based model to predict surgical wait times in paediatric oncology patients. Using real-world data from 1478 patients and 6145 surgeries, the model first classifies surgical urgency, the...This study presents a two-phase AI-based model to predict surgical wait times in paediatric oncology patients. Using real-world data from 1478 patients and 6145 surgeries, the model first classifies surgical urgency, then estimates wait times for urgent cases. Random Forest emerged as the best-performing algorithm in both phases, and SHAP analysis identified similar key predictive features. Results support AI's role in improving surgical planning, resource allocation, and clinical decision-making.
Recenti Prog Med
· 2025 Oct · PMID 41037374
·
Publisher ↗
AI is transforming neurology, providing powerful diagnostic and therapeutic tools, yet handling sensitive clinical data involves substantial privacy risks. Federated Learning addresses these issues by training models loc...AI is transforming neurology, providing powerful diagnostic and therapeutic tools, yet handling sensitive clinical data involves substantial privacy risks. Federated Learning addresses these issues by training models locally within hospitals and sharing only their weights or gradients for final training, achieving similar or superior performance compared to centralized models in Stroke, Alzheimer's, and Parkinson's disease.
Recenti Prog Med
· 2025 Oct · PMID 41037373
·
Publisher ↗
This study offers a comparative evaluation of fifteen generative artificial intelligence models using 397 Italian multiple-choice questions on occupational medicine. Model accuracy ranged from 75.06% to 95.72%. The resul...This study offers a comparative evaluation of fifteen generative artificial intelligence models using 397 Italian multiple-choice questions on occupational medicine. Model accuracy ranged from 75.06% to 95.72%. The results highlight the need to assess large language models in specialized fields to support their safe and effective integration into medical education and occupational medicine practice.
Traverso A, Barbieri S, Denti M
… +2 more, Esposito A, Tacchetti C
Recenti Prog Med
· 2025 Oct · PMID 41037372
·
Publisher ↗
The S-RACE platform is a cloud-based AI solution for using real-world health data. It addresses data quality and governance challenges with an end-to-end pipeline, including on-premise anonymisation and tools for clinici...The S-RACE platform is a cloud-based AI solution for using real-world health data. It addresses data quality and governance challenges with an end-to-end pipeline, including on-premise anonymisation and tools for clinicians and data scientists. Aligned with responsible AI principles, it aims to accelerate the translation of AI research into clinical practice, improving patient care.
Mondillo G, Perrotta A, Frattolillo V
… +4 more, Masino M, Colosimo S, Abbate FG, Marzuillo P
Recenti Prog Med
· 2025 Oct · PMID 41037371
·
Publisher ↗
Model Context Protocol (MCP) is an open standard for connecting AI applications to external tools. While not designed for healthcare, it offers advantages for clinical decision support through natural language queries. W...Model Context Protocol (MCP) is an open standard for connecting AI applications to external tools. While not designed for healthcare, it offers advantages for clinical decision support through natural language queries. We developed a pediatric MCP server with 46 clinical tools and tested it using 32 cases. Results: correctly processed 31/32 cases. This is the first clinical validation of MCP technology, demonstrating high reliability for clinical application.
Recenti Prog Med
· 2025 Oct · PMID 41037370
·
Publisher ↗
DermatAI is the AI-based tool developed for early skin cancer detection, with a focus on Melanoma. Through an ensemble stacking model composed by convolutional neural networks and XGBoost classifier. DermatAI classifies...DermatAI is the AI-based tool developed for early skin cancer detection, with a focus on Melanoma. Through an ensemble stacking model composed by convolutional neural networks and XGBoost classifier. DermatAI classifies skin lesions with high accuracy, aiding melanoma diagnosis and improving clinical decision support.
Montomoli J, Iannaccone S, Russo S
… +6 more, Coletta A, Cappucci O, Folla M, Placidi V, Frontoni E, Giuliani F
Recenti Prog Med
· 2025 Oct · PMID 41037369
·
Publisher ↗
The MedWriter project aims to create an AI-based clinical decision support system for automated discharge letter generation. Co-funded through Italian Sustainable Growth Fund and the EU, the project will utilize 1.3M pat...The MedWriter project aims to create an AI-based clinical decision support system for automated discharge letter generation. Co-funded through Italian Sustainable Growth Fund and the EU, the project will utilize 1.3M patient records from HL7/FHIR-compliant SISWEB platform. The system will employ hybrid neural architectures including CNNs, transformers, and reinforcement learning for accurate clinical narrative synthesis.
Roveta A, Castello LM, Ugo F
… +5 more, Petronio M, Terenziani P, Bottrighi A, Raina E, Maconi A
Recenti Prog Med
· 2025 Oct · PMID 41037368
·
Publisher ↗
GLARE-Edu is an AI-powered, adaptive platform supporting healthcare professionals and students in learning clinical guidelines and improving decision-making through personalized training and realistic case simulations. T...GLARE-Edu is an AI-powered, adaptive platform supporting healthcare professionals and students in learning clinical guidelines and improving decision-making through personalized training and realistic case simulations. Two case studies demonstrated significant improvements in guideline application and user satisfaction.
Panizzi M, Bellini V, Domenichetti T
… +3 more, Darhour L, Sancricca L, Bignami E
Recenti Prog Med
· 2025 Oct · PMID 41037367
·
Publisher ↗
The "Insieme" project applies the metaverse to anesthesia by creating a digital twin of the Operating Suite to improve the doctor-patient relationship, reduce preoperative anxiety, and support continuous education. Patie...The "Insieme" project applies the metaverse to anesthesia by creating a digital twin of the Operating Suite to improve the doctor-patient relationship, reduce preoperative anxiety, and support continuous education. Patients undergo an immersive virtual journey, interacting with nurse and doctor avatars, simulating the surgical experience and postoperative awakening.
Bellini V, Panizzi M, Domenichetti T
… +2 more, Guarnieri M, Bignami E
Recenti Prog Med
· 2025 Oct · PMID 41037366
·
Publisher ↗
The combined use of IoT and AI enables automatic and precise collection of operative times through BLE bracelets, improving efficiency compared to manual recording. Surgery-specific models, trained on real data, better p...The combined use of IoT and AI enables automatic and precise collection of operative times through BLE bracelets, improving efficiency compared to manual recording. Surgery-specific models, trained on real data, better predict procedure duration, optimizing management and resources in the operating room.
Ricci G, Capuzzi S, Riganati M
… +4 more, Tozzi AE, Vivarelli M, Ferro D, Colucci M
Recenti Prog Med
· 2025 Oct · PMID 41037365
·
Publisher ↗
This study explores lymphocyte profiles as non-invasive biomarkers for classification of pediatric nephrotic syndrome (NS). Using retrospective clinical and immunological data from 205 patients, the aim is to develop a p...This study explores lymphocyte profiles as non-invasive biomarkers for classification of pediatric nephrotic syndrome (NS). Using retrospective clinical and immunological data from 205 patients, the aim is to develop a predictive model based on Long Short-Term Memory to identify NS subtypes. By comparing models with and without immunological data, the study will assess the value of immune profiles. The goal is to support personalized management while reducing the need for invasive procedures.
Marchi G, Gambini G, Guglielmi G
… +2 more, Pistelli F, Carrozzi L
Recenti Prog Med
· 2025 Oct · PMID 41037364
·
Publisher ↗
Three LLMs - ChatGPT-4, Claude 3.5 Sonnet and Gemini 1.5 Advanced - were evaluated on COPD questions from the GOLD recommendations. Sixty-one pulmonologists from 6 continents rated 90 AI responses for completeness, accur...Three LLMs - ChatGPT-4, Claude 3.5 Sonnet and Gemini 1.5 Advanced - were evaluated on COPD questions from the GOLD recommendations. Sixty-one pulmonologists from 6 continents rated 90 AI responses for completeness, accuracy, terminology, accessibility, and safety. Gemini outperformed in completeness, Claude in accuracy and terminology, with no differences in accessibility or safety. While promising, clinical use requires caution and further validation to ensure safe, accurate patient education.
Giuliani F, Cappucci O, De Gennaro C
… +6 more, Ricciardi F, Russo S, Copetti M, Crociani P, Pugliatti M, Leone M
Recenti Prog Med
· 2025 Oct · PMID 41037363
·
Publisher ↗
The study evaluated the use of a popular large language model (LLM) to support neurologists in communicating with patients with Multiple Sclerosis. We describe the development of a tailored COSTAR prompt and the process...The study evaluated the use of a popular large language model (LLM) to support neurologists in communicating with patients with Multiple Sclerosis. We describe the development of a tailored COSTAR prompt and the process that led to its refinement. A cohort of neurologists assessed the prompt's effectiveness using the QAMAI tool. The results highlight both strengths and the issues that must be addressed for the effective clinical use of LLMs in this context.
Recenti Prog Med
· 2025 Oct · PMID 41037362
·
Publisher ↗
Breakthrough digital phenotyping approach reveals three distinct behavioral patterns from smartphone data that could revolutionize personalized mental health care. Using AI clustering on 77 users, we discovered "Night Ow...Breakthrough digital phenotyping approach reveals three distinct behavioral patterns from smartphone data that could revolutionize personalized mental health care. Using AI clustering on 77 users, we discovered "Night Owls", "Routine-Oriented", and "Always-Connected" behavioral types with 90%+ accuracy. Our explainable ML pipeline identifies key digital biomarkers for targeted interventions, offering clinicians data-driven insights for precision psychiatry.
Ferro D, Baglivo F, De Angelis L
… +7 more, Causio FA, Di Pumpo M, Sacchi FA, Diedenhofen G, Pivetta A, Belpiede A, Tozzi AE
Recenti Prog Med
· 2025 Oct · PMID 41037361
·
Publisher ↗
Precision medicine seeks to tailor care by integrating genetic, clinical, and environmental data. Digital twins, dynamic, virtual replicas of patients that are updated with longitudinal information, represent a significa...Precision medicine seeks to tailor care by integrating genetic, clinical, and environmental data. Digital twins, dynamic, virtual replicas of patients that are updated with longitudinal information, represent a significant step in this direction. Enabled by artificial intelligence, they allow in silico experimentation to simulate therapies, disease trajectories, and adverse events, reducing risk and sharpening personalization. By bridging data and decisions, digital twins can promote earlier diagnosis, targeted treatments, and faster drug discovery, supporting a shift from reactive to predictive and participatory care. Nonetheless, challenges surrounding data integration, privacy, regulation, and equity persist and necessitate collaborative solutions. This viewpoint examines the opportunities and system-level requirements to integrate digital twins into Italian healthcare. Digital twins redefine medicine by turning episodic encounters into continuous, adaptive care. They can anticipate clinical events, simulate individualized treatments, and support shared decision-making, advancing the vision of predictive, preventive, personalized, and participatory medicine. Realizing this potential requires robust governance, interoperable infrastructures, and clinician training, alongside ethical frameworks that protect autonomy and fairness. Public-private partnerships and international collaboration will be crucial for the responsible, inclusive, and transparent adoption of these initiatives. Ultimately, digital twins inaugurate a paradigm in which simulation and clinical reality converge, fostering innovation that is both scientifically rigorous and deeply human.
De Angelis L, Pivetta A, Baglivo F
… +10 more, Cappellini LA, Sacchi FA, Di Pumpo M, Mercier M, Diedenhofen G, Di Bartolomeo M, Causio FA, Belpiede A, Tozzi AE, Ferro D
Recenti Prog Med
· 2025 Oct · PMID 41037360
·
Publisher ↗
In Italy, the growing enthusiasm for artificial intelligence (AI) in healthcare contrasts with significant infrastructural, cultural, and trust-related barriers hindering its real-world adoption. Moving beyond the hype r...In Italy, the growing enthusiasm for artificial intelligence (AI) in healthcare contrasts with significant infrastructural, cultural, and trust-related barriers hindering its real-world adoption. Moving beyond the hype requires a systems thinking approach, proposing the learning health system (LHS) framework as a structured path for integration. We highlight the complementary roles of AI models: traditional machine learning (ML) is proven for diagnostics and prognostics, while large language models (LLMs) excel at administrative tasks and can structure unstructured data to train robust ML tools. The LHS cycle reveals key challenges for Italy: moving from Practice-to-Data requires overcoming data fragmentation; from Data-to-Knowledge involves transforming data into insights while mitigating bias; and from Knowledge-to-Practice necessitates bridging the gap between evidence and clinical workflow by building trust and AI literacy. Ultimately, successful and equitable AI implementation depends on a holistic strategy combining infrastructure development, multidisciplinary collaboration, and robust governance to enhance the quality and sustainability of the national healthcare system.