Ioana-Alexandra Stoia
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Ioana-Alexandra Stoia has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Claudia-Mihaela Păun
Iuliu Hatieganu University of Medicine and Pharmacy
4 shared publications - Victor Volovici
Department of Neurosurgery, Erasmus MC Stroke Center, Erasmus MC University Medical Center, Rotterdam, The Netherlands
4 shared publications - Alexandru Ilie-Ene
Iuliu Hatieganu University of Medicine and Pharmacy
3 shared publications - George Calin Dindelegan
Department of General Surgery, “Iuliu Hatieganu” University of Medicine and Pharmacy, 400012 Cluj-Napoca, Romania
3 shared publications - Victor Esanu
Center for Complex Microvascular Surgery, Erasmus MC University Medical Center, 3015 GD Rotterdam, The Netherlands
3 shared publications - Clemens Maria Franciscus Dirven
Department of Neurosurgery, Erasmus MC University Medical Center, 3015 GD Rotterdam, The Netherlands
2 shared publications
Latest Publications
- 10 November 2026
- Published by Springer Nature
- p. 109-115
Vena Cava End-to-End Anastomosis
- 10 November 2026
- Published by Springer Nature
- p. 7-14
Microsurgical Setup
- 10 November 2026
- Published by Springer Nature
- p. 67-79
Rat Retroperitoneum Dissection and Aorta End-to-End Anastomosis
- 10 November 2026
- Published by Springer Nature
- p. 67-79
Rat Retroperitoneum Dissection and Aorta End-to-End Anastomosis
- 10 November 2026
- Published by Springer Nature
- p. 169-177
Psychomotor Skill Assessment in Microsurgery
- Open Access
- 2 May 2026
- Published by MDPI AG in Medical Sciences
- Vol. 14 (2) , 235
Vision-Capable LLMs in Microsurgery: A Blinded Comparison of Two AI Models with Expert Microsurgeons in the Appraisal of 200 Experimental Anastomoses
Background/Objectives: Objective end-product assessment of microsurgical anastomoses is intensive and partly subjective. Vision-capable large language models (LLMs) may enable standardized image-based scoring, but their agreement with expert assessment remains uncertain. Methods: We studied 200 end-to-end femoral artery anastomoses, performed on chicken legs by novice, intermediate, and experienced microsurgeons. Images were scored independently by two blinded expert panels; disagreements were adjudicated by a third senior reviewer to establish expert consensus. Two LLMs, ChatGPT 5.2 Thinking Extended and Gemini 3.1 Pro, were evaluated using the exact same prompt and rubric. Each image was analyzed three times per model. Final scores were aggregated by median for numeric items and majority vote for categorical items. The primary endpoint was exact-match agreement with expert consensus. Agreement within ±1 was also assessed for numeric items. Agreement was measured using simple percentage agreement, Light’s kappa, and Krippendorff’s alpha; Bland–Altman analysis was used for numeric count items. Results: LLM 1 achieved a higher overall exact-match agreement than LLM 2 (0.659 vs. 0.539). Both models performed better on categorical than numeric items (0.713 vs. 0.610 and 0.651 vs. 0.445, respectively). LLM 1 showed the greatest advantages for gaps, knots, oblique stitches, and wide bites. Krippendorff’s alpha was positive for most endpoints with LLM 1, whereas LLM 2 showed negative values throughout. Allowing a ±1 tolerance for numeric items greatly improved agreement, suggesting only minor counting discrepancies, from 0.610 to 0.900 for LLM 1 and from 0.445 to 0.826 for LLM 2. Conclusions: Under a constrained scoring workflow, LLMs partially approximated intraluminal microsurgical end-product scoring. LLM 1 outperformed LLM 2, but agreement remained insufficient to replace the expert assessment entirely. These models can be assistive tools within a human-in-the-loop framework.
- 2 May 2026
- Published by MDPI AG in Medical Sciences
- Vol. 14 (2) , 235
Open AccessVision-Capable LLMs in Microsurgery: A Blinded Comparison of Two AI Models with Expert Microsurgeons in the Appraisal of 200 Experimental Anastomoses
Background/Objectives: Objective end-product assessment of microsurgical anastomoses is intensive and partly subjective. Vision-capable large language models (LLMs) may enable standardized image-based scoring, but their agreement with expert assessment remains uncertain. Methods: We studied 200 end-to-end femoral artery anastomoses, performed on chicken legs by novice, intermediate, and experienced microsurgeons. Images were scored independently by two blinded expert panels; disagreements were adjudicated by a third senior reviewer to establish expert consensus. Two LLMs, ChatGPT 5.2 Thinking Extended and Gemini 3.1 Pro, were evaluated using the exact same prompt and rubric. Each image was analyzed three times per model. Final scores were aggregated by median for numeric items and majority vote for categorical items. The primary endpoint was exact-match agreement with expert consensus. Agreement within ±1 was also assessed for numeric items. Agreement was measured using simple percentage agreement, Light’s kappa, and Krippendorff’s alpha; Bland–Altman analysis was used for numeric count items. Results: LLM 1 achieved a higher overall exact-match agreement than LLM 2 (0.659 vs. 0.539). Both models performed better on categorical than numeric items (0.713 vs. 0.610 and 0.651 vs. 0.445, respectively). LLM 1 showed the greatest advantages for gaps, knots, oblique stitches, and wide bites. Krippendorff’s alpha was positive for most endpoints with LLM 1, whereas LLM 2 showed negative values throughout. Allowing a ±1 tolerance for numeric items greatly improved agreement, suggesting only minor counting discrepancies, from 0.610 to 0.900 for LLM 1 and from 0.445 to 0.826 for LLM 2. Conclusions: Under a constrained scoring workflow, LLMs partially approximated intraluminal microsurgical end-product scoring. LLM 1 outperformed LLM 2, but agreement remained insufficient to replace the expert assessment entirely. These models can be assistive tools within a human-in-the-loop framework.
- Open Access
- 22 October 2025
- Published by MDPI AG in Journal of Clinical Medicine
- Vol. 14 (21) , 7452
Validated Microsurgical Training Programmes: A Systematic Review of the Current Literature
Background: Microsurgical skill acquisition and development are complex processes, due to the often complex learning curve, limited training possibilities, and growing restrictions on working hours. Simulation-based training programmes, employing various models, have been proposed. Nevertheless, the extent to which these training programmes are supported by scientific evidence is unclear. The aim of this systematic review is to evaluate the extent and quality of the scientific evidence backing validated microsurgical training programmes. Methods: A systematic literature review was conducted, following a study protocol established a priori and in accordance with the PRISMA guidelines. The databases searched were the Web of Science Core Collection (Web of Knowledge), Medline (Ovid), Embase (Embase.com), and ERIC (Ovid). Studies were included if they described microsurgical training programmes and presented a form of validation of training effectiveness. Data extraction included the number of participants, training duration and frequency, validation type, assessment methods, outcomes, study limitations, and a detailed training regimen. The risk of bias and quality were assessed using the Medical Education Research Study Quality Instrument (MERSQI). Validity was assessed using an established validity framework (content, face, construct, and criterion encompassing both concurrent and predictive validity). The Level of Evidence (LoE) and Recommendation (LoR) were evaluated using the Oxford Centre for Evidence-Based Medicine (OCEBM). Results: A total of 25 studies met the inclusion criteria. Training programmes were classified into one-time intensive courses or longitudinal curricula. Face, content, and construct validity were the most commonly assessed aspects, while predictive validity was the least frequently assessed and not properly evaluated. Training models ranged from low-fidelity (silicone tubes, synthetic vessels) to high-fidelity (live animal models). The Global Rating Scale (GRS), the Structured Assessment of Microsurgery Skills (SAMS), and the Objective Structured Assessment of Technical Skills (OSATS) were the most frequently used objective assessment tools for evaluation methods within the programmes. The risk of bias MERSQI score was 12.96, ranging from 10.5 to 15.5, and LoE and LoR scores were moderated. Across the studies, 96% reported significant improvement in microsurgical skills among participants. However, most studies were limited by small sample sizes, heterogeneity in baseline skills, and a lack of long-term follow-up. Conclusions: While validated microsurgical training programmes improve skill acquisition, challenges remain in terms of standardisation and best cost-effective methods. Future research should prioritise evaluating predictive validity, creating standardised objective assessment tools, and focus on skill maintenance.
- 22 October 2025
- Published by MDPI AG in Journal of Clinical Medicine
- Vol. 14 (21) , 7452
Open AccessValidated Microsurgical Training Programmes: A Systematic Review of the Current Literature
Background: Microsurgical skill acquisition and development are complex processes, due to the often complex learning curve, limited training possibilities, and growing restrictions on working hours. Simulation-based training programmes, employing various models, have been proposed. Nevertheless, the extent to which these training programmes are supported by scientific evidence is unclear. The aim of this systematic review is to evaluate the extent and quality of the scientific evidence backing validated microsurgical training programmes. Methods: A systematic literature review was conducted, following a study protocol established a priori and in accordance with the PRISMA guidelines. The databases searched were the Web of Science Core Collection (Web of Knowledge), Medline (Ovid), Embase (Embase.com), and ERIC (Ovid). Studies were included if they described microsurgical training programmes and presented a form of validation of training effectiveness. Data extraction included the number of participants, training duration and frequency, validation type, assessment methods, outcomes, study limitations, and a detailed training regimen. The risk of bias and quality were assessed using the Medical Education Research Study Quality Instrument (MERSQI). Validity was assessed using an established validity framework (content, face, construct, and criterion encompassing both concurrent and predictive validity). The Level of Evidence (LoE) and Recommendation (LoR) were evaluated using the Oxford Centre for Evidence-Based Medicine (OCEBM). Results: A total of 25 studies met the inclusion criteria. Training programmes were classified into one-time intensive courses or longitudinal curricula. Face, content, and construct validity were the most commonly assessed aspects, while predictive validity was the least frequently assessed and not properly evaluated. Training models ranged from low-fidelity (silicone tubes, synthetic vessels) to high-fidelity (live animal models). The Global Rating Scale (GRS), the Structured Assessment of Microsurgery Skills (SAMS), and the Objective Structured Assessment of Technical Skills (OSATS) were the most frequently used objective assessment tools for evaluation methods within the programmes. The risk of bias MERSQI score was 12.96, ranging from 10.5 to 15.5, and LoE and LoR scores were moderated. Across the studies, 96% reported significant improvement in microsurgical skills among participants. However, most studies were limited by small sample sizes, heterogeneity in baseline skills, and a lack of long-term follow-up. Conclusions: While validated microsurgical training programmes improve skill acquisition, challenges remain in terms of standardisation and best cost-effective methods. Future research should prioritise evaluating predictive validity, creating standardised objective assessment tools, and focus on skill maintenance.
- 7 July 2022
- Published by Georg Thieme Verlag KG in Journal of Reconstructive Microsurgery
- Vol. 38 (08) , 604-612
Reduction of the Number of Live Animals Used for Microsurgical Skill Acquisition: An Experimental Randomized Noninferiority Trial
Background Live animals have been used for decades as one of the many training models for developing surgical skills. Microsurgery in particular relies on training for skill acquisition and maintenance, using live animal models, especially rats (murine models). Efforts are underway to reduce the number of rats sacrificed to achieve this objective. Methods A group of students with minimal microsurgical experience, after having gone through a basic microsurgical course, were randomly split into three equal groups, all three groups following a 24-week standard training program based on low- and medium-fidelity training models with standardized murine training days. In addition to the standard training regimen, each participant performed supplementary training on live rats every 4, 6, or 8 weeks. According to the training program, the procedures have been performed on chicken legs, flower petals, and rats, each procedure being blindly assessed and evaluated using validated models and scales. The primary evaluated outcome was the Stanford Microsurgery and Resident Training (SMaRT) scale result of the final rat anastomosis performed by each group, for which the tested hypothesis was one of noninferiority. The secondary outcomes were represented by the final rat anastomosis time, final chicken leg anastomosis result and time, and the final petal score. Results After the 24th week, no differences were observed between the three groups regarding their microsurgical skills, as measured by the aforementioned surgical outcomes. All participants improved significantly during the study (mean [standard deviation] 19 ± 4 points on the SMaRT scale), with no significant differences between the groups, p < 0.001 for noninferiority. Conclusion A training regimen based on low- and moderate-fidelity models, with the addition of training on a live rat every 8 weeks was noninferior to a training regimen that used a live rat every 6 weeks and also noninferior to a training regimen that used a live rat every 4 weeks.
- 7 July 2022
- Published by Georg Thieme Verlag KG in Journal of Reconstructive Microsurgery
- Vol. 38 (08) , 604-612
Reduction of the Number of Live Animals Used for Microsurgical Skill Acquisition: An Experimental Randomized Noninferiority Trial
Background Live animals have been used for decades as one of the many training models for developing surgical skills. Microsurgery in particular relies on training for skill acquisition and maintenance, using live animal models, especially rats (murine models). Efforts are underway to reduce the number of rats sacrificed to achieve this objective. Methods A group of students with minimal microsurgical experience, after having gone through a basic microsurgical course, were randomly split into three equal groups, all three groups following a 24-week standard training program based on low- and medium-fidelity training models with standardized murine training days. In addition to the standard training regimen, each participant performed supplementary training on live rats every 4, 6, or 8 weeks. According to the training program, the procedures have been performed on chicken legs, flower petals, and rats, each procedure being blindly assessed and evaluated using validated models and scales. The primary evaluated outcome was the Stanford Microsurgery and Resident Training (SMaRT) scale result of the final rat anastomosis performed by each group, for which the tested hypothesis was one of noninferiority. The secondary outcomes were represented by the final rat anastomosis time, final chicken leg anastomosis result and time, and the final petal score. Results After the 24th week, no differences were observed between the three groups regarding their microsurgical skills, as measured by the aforementioned surgical outcomes. All participants improved significantly during the study (mean [standard deviation] 19 ± 4 points on the SMaRT scale), with no significant differences between the groups, p < 0.001 for noninferiority. Conclusion A training regimen based on low- and moderate-fidelity models, with the addition of training on a live rat every 8 weeks was noninferior to a training regimen that used a live rat every 6 weeks and also noninferior to a training regimen that used a live rat every 4 weeks.