Nicola Osti
Nicola Osti has a total of 37 co-authors. Here are the top co-authors by number of shared publications:
- Alberto Maino
Department of Clinical Epidemiology, Leiden University Medical Center, Leiden, The Netherlands
2 shared publications - Nicola Susca
Internal Medicine Division, Santa Chiara Hospital, Azienda Sanitaria Universitaria Integrata del Trentino (ASUIT), Trento, Italy
2 shared publications - Vito Racanelli
Center for Medical Sciences (CISMed), University of Trento, Trento, Italy
2 shared publications Adriano Anesi
Clinical Pathology Laboratory, Laboratories Department, ASUIT Trento, (Brocco and Anesi), Trento, Italy
1 shared publication- Andrea Passerini
University of Trento, Trento, Italy
1 shared publication - Anna Brugnolli
Center for Medical Sciences (CISMed), University of Trento, Trento, Italy
1 shared publication
Latest Publications
- Open Access
- 12 June 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Human–LLM Dialogue Improves Diagnostic Accuracy in Emergency Care: Controlled Crossover Study (Preprint)
BACKGROUND Clinical decision-making in emergency medicine requires rapid, accurate diagnoses under uncertainty. Large language models (LLMs) have shown diagnostic potential, but evidence for their benefit in real physician workflows remains limited. OBJECTIVE This study evaluated whether MedSyn, a physician–LLM dialogue system, improves diagnostic accuracy in live emergency care workflows, and examined how benefit varies with case difficulty and physician expertise. METHODS Seven physicians (three senior internists, four residents) completed unassisted (baseline) and AI-assisted (interactive) sessions across 52 MIMIC-IV cases stratified by difficulty (Easy, Medium, Hard). Diagnostic performance was measured using automated fuzzy-matching metrics (any-match accuracy, exact-match accuracy, diagnosis-set F1) and blinded expert evaluation. Between-condition comparisons used paired bootstrap resampling (20,000 replicates, seed 42). RESULTS Interactive support significantly improved overall diagnostic accuracy: difficulty-standardised any-match accuracy improved by 0.156 (P < .001; d = 1.25). Residents showed the largest gains, with F1 improving by 0.138 (P < .001; d = 0.94). Blinded evaluation revealed residents’ Hard-case mean correctness rose from 0.589 to 0.734 (+0.145); difficulty-standardised completely-correct rates confirmed a medium effect (Δ = 0.092; P = .07; d = 0.47). Automated and blinded manual assessment showed only modest agreement (Cohen's κ = 0.148), with 45% of clinically correct diagnoses receiving no automated credit; reported F1 improvements are therefore conservative estimates. Cross-expertise concordance increased significantly (Δ = 0.142; P < .001; d = 1.48). Dialogue analysis revealed expertise-dependent strategies: seniors employed targeted, hypothesis-driven queries while residents used broader exploratory questioning. CONCLUSIONS Interactive LLM assistance meaningfully enhances diagnostic accuracy, particularly for resident physicians on complex cases. The expertise-stratified benefit profile, large effect sizes, and significant convergence in cross-expertise concordance suggest that adaptive LLM deployment, targeting high-complexity cases and less experienced clinicians, is a promising and evidence-grounded strategy for emergency diagnostic support.
- 12 June 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessHuman–LLM Dialogue Improves Diagnostic Accuracy in Emergency Care: Controlled Crossover Study (Preprint)
BACKGROUND Clinical decision-making in emergency medicine requires rapid, accurate diagnoses under uncertainty. Large language models (LLMs) have shown diagnostic potential, but evidence for their benefit in real physician workflows remains limited. OBJECTIVE This study evaluated whether MedSyn, a physician–LLM dialogue system, improves diagnostic accuracy in live emergency care workflows, and examined how benefit varies with case difficulty and physician expertise. METHODS Seven physicians (three senior internists, four residents) completed unassisted (baseline) and AI-assisted (interactive) sessions across 52 MIMIC-IV cases stratified by difficulty (Easy, Medium, Hard). Diagnostic performance was measured using automated fuzzy-matching metrics (any-match accuracy, exact-match accuracy, diagnosis-set F1) and blinded expert evaluation. Between-condition comparisons used paired bootstrap resampling (20,000 replicates, seed 42). RESULTS Interactive support significantly improved overall diagnostic accuracy: difficulty-standardised any-match accuracy improved by 0.156 (P < .001; d = 1.25). Residents showed the largest gains, with F1 improving by 0.138 (P < .001; d = 0.94). Blinded evaluation revealed residents’ Hard-case mean correctness rose from 0.589 to 0.734 (+0.145); difficulty-standardised completely-correct rates confirmed a medium effect (Δ = 0.092; P = .07; d = 0.47). Automated and blinded manual assessment showed only modest agreement (Cohen's κ = 0.148), with 45% of clinically correct diagnoses receiving no automated credit; reported F1 improvements are therefore conservative estimates. Cross-expertise concordance increased significantly (Δ = 0.142; P < .001; d = 1.48). Dialogue analysis revealed expertise-dependent strategies: seniors employed targeted, hypothesis-driven queries while residents used broader exploratory questioning. CONCLUSIONS Interactive LLM assistance meaningfully enhances diagnostic accuracy, particularly for resident physicians on complex cases. The expertise-stratified benefit profile, large effect sizes, and significant convergence in cross-expertise concordance suggest that adaptive LLM deployment, targeting high-complexity cases and less experienced clinicians, is a promising and evidence-grounded strategy for emergency diagnostic support.
- Open Access
- 5 June 2026
- Published by Elsevier in Journal of Clinical Lipidology
- Vol. 20 (8) , 1605-1615
Lipoprotein(a) plasma levels and coagulation biomarkers: Results from a comprehensive laboratory assessment in an angiographically controlled cardiovascular cohort
- 5 June 2026
- Published by Elsevier in Journal of Clinical Lipidology
- Vol. 20 (8) , 1605-1615
Open AccessLipoprotein(a) plasma levels and coagulation biomarkers: Results from a comprehensive laboratory assessment in an angiographically controlled cardiovascular cohort
- Open Access
- 1 January 2025
- Published by Wiley in Emergency Medicine International
- Vol. 2025 (1) , 2807776
Implementation of Revised Simplified Geneva Score in Triage Nurse Evaluation for Patients With Suspected Pulmonary Embolism: A Retrospective Chart Review
Background: Pulmonary embolism (PE) is a clinical condition frequently encountered in the emergency department (ED), with a high and early mortality rate. ED triage determines the priority of further evaluation of care at the time of patient arrival. Very little is known about the specific role of ED triage in PE. We aimed to evaluate (1) whether the current five‐level triage (5LT) system can identify patients with PE and differently prioritize them for medical evaluation and (2) the discriminatory capacity of simplified revised Geneva score (SRGS) toward PE diagnosis when calculated by triage nurses. Methods: A retrospective chart review on ED patients who underwent computed tomography pulmonary angiography (CTPA) in 2023. Based on the CTPA report, patients were categorized into two subgroups: CTPA PE‐negative and CTPA PE‐positive. We then searched for correlations between PE diagnosis and triage priority level, time from triage to medical evaluation, SRGS, and National Early Warning Score 2 (NEWS2). Results: Of the 196 patients included in the analysis (age 71.1 ± 16.9), 45 (23.0%) were CTPA PE‐positive (26 proximal PE and 19 distal PE). There was no correlation between the assigned triage color code and the CTPA results. Although we found a statistically significant difference in the prevalence of CTPA‐confirmed PE according to the results of the SRGS ( p = 0.014), the SRGS calculated at the time of triage showed a poor prediction accuracy for subsequent PE diagnosis (area under curve [AUC] 0.608). NEWS2 was significantly associated with the triage‐assigned priority level ( p < 0.001). Conclusions: The current 5LT was unable to differently prioritize patients with or without PE, and it seems unlikely that implementation of SRGS in the triage nurse evaluation will significantly improve the prioritization of patients with suspected PE for medical evaluation. Nonetheless, application of SRGS in triage evaluation may improve the appropriateness of the subsequent clinical pathway for PE diagnosis and risk stratification.
- 1 January 2025
- Published by Wiley in Emergency Medicine International
- Vol. 2025 (1) , 2807776
Open AccessImplementation of Revised Simplified Geneva Score in Triage Nurse Evaluation for Patients With Suspected Pulmonary Embolism: A Retrospective Chart Review
Background: Pulmonary embolism (PE) is a clinical condition frequently encountered in the emergency department (ED), with a high and early mortality rate. ED triage determines the priority of further evaluation of care at the time of patient arrival. Very little is known about the specific role of ED triage in PE. We aimed to evaluate (1) whether the current five‐level triage (5LT) system can identify patients with PE and differently prioritize them for medical evaluation and (2) the discriminatory capacity of simplified revised Geneva score (SRGS) toward PE diagnosis when calculated by triage nurses. Methods: A retrospective chart review on ED patients who underwent computed tomography pulmonary angiography (CTPA) in 2023. Based on the CTPA report, patients were categorized into two subgroups: CTPA PE‐negative and CTPA PE‐positive. We then searched for correlations between PE diagnosis and triage priority level, time from triage to medical evaluation, SRGS, and National Early Warning Score 2 (NEWS2). Results: Of the 196 patients included in the analysis (age 71.1 ± 16.9), 45 (23.0%) were CTPA PE‐positive (26 proximal PE and 19 distal PE). There was no correlation between the assigned triage color code and the CTPA results. Although we found a statistically significant difference in the prevalence of CTPA‐confirmed PE according to the results of the SRGS ( p = 0.014), the SRGS calculated at the time of triage showed a poor prediction accuracy for subsequent PE diagnosis (area under curve [AUC] 0.608). NEWS2 was significantly associated with the triage‐assigned priority level ( p < 0.001). Conclusions: The current 5LT was unable to differently prioritize patients with or without PE, and it seems unlikely that implementation of SRGS in the triage nurse evaluation will significantly improve the prioritization of patients with suspected PE for medical evaluation. Nonetheless, application of SRGS in triage evaluation may improve the appropriateness of the subsequent clinical pathway for PE diagnosis and risk stratification.