Ignacio Villagran

Plain English
Physiotherapy students learned airway suctioning technique through a four-week remote simulation program using a 3D-printed airway model and video-based instructor feedback, without any in-person contact. By the end of training, students' procedural scores matched expert performance on standardized assessment. This shows remote simulation can reliably build competency in a high-stakes clinical procedure.

Plain English
Researchers co-designed a feedback toolkit with clinical teachers and students for use during physiotherapy internships, testing it over seven weeks. The toolkit—built around podcasts, infographics, and Mini-CEX forms—helped structure ongoing feedback between teachers and students in a more consistent way. The process demonstrated that involving end users directly in designing educational tools produces materials that are actually used in practice.

Plain English
GPT-4o was tested on 183 questions from Chile's national anesthesiology board exam across 30 simulation runs. It scored 83.7% overall, with strongest performance on recall and comprehension questions but weaker performance on application and analysis tasks. The most common errors were unsupported medical claims, highlighting that AI systems still struggle with complex clinical reasoning before they can be trusted in high-stakes settings.

Plain English
Medical students performed simulated paracentesis procedures, received instructor feedback scored by an AI tool called Teach1, then attempted the procedure again. Higher AI-rated feedback quality strongly predicted larger skill improvements, with each 10% increase in feedback quality corresponding to a 1.16-point score gain. This proves that feedback quality—not just quantity—drives skill development, and that AI can reliably identify which feedback is actually effective.

Plain English
This paper provides a practical guide for educators wanting to use multimodal assessment in simulation-based medical training, where data from multiple sources (video, sensors, performance metrics) are combined to evaluate learners. The guide offers a structured checklist to help researchers plan these assessments and anticipate common implementation pitfalls. Combining data streams gives a richer picture of learner performance than any single measure alone.

Plain English
GPT-4o was tested on 1,062 multiple-choice questions from Chile's medical licensing exam using both single-agent and multi-agent AI strategies. The multi-agent approach called MDAGENTS achieved the highest accuracy at nearly 90%, significantly outperforming simpler strategies. For most exam questions, however, simpler single-agent methods were sufficient—meaning complex multi-agent setups add meaningful value only for the hardest questions.

Plain English
Researchers adapted and validated a Spanish-language version of the Feedback Orientation Scale for use with 510 health professions students at six Chilean universities. The scale reliably measured how students perceive and use feedback in clinical training, with the lowest scores appearing on questions about confidence in handling feedback. This validated tool lets educators identify which students may struggle to act on feedback and tailor interventions accordingly.

Plain English
Third-year medical students were trained in thoracentesis and paracentesis using either an unsupervised remote simulation program with asynchronous video feedback or a two-hour in-person workshop. Students in the remote group achieved significantly higher passing rates on both procedures (80% vs. 43% for thoracentesis; 91% vs. 67% for paracentesis). Asynchronous feedback allows students and instructors to work at their own pace without sacrificing—and in this case improving—skill acquisition.

Plain English
The study developed a low-cost, DIY ultrasound-compatible wrist phantom for training joint aspiration using 3D-printed bones embedded in gelatin. The model accurately reproduces the ultrasound appearance of wrist anatomy and can be remade after each practice session to eliminate needle tracks. This gives clinical simulation centers a replicable, affordable alternative to expensive commercial trainers for a procedure that is performed infrequently.

Plain English
A scoping review of 61 studies examined how feedback is organized in undergraduate clinical training settings. Multiple teaching and assessment activities generate feedback, but very few studies tracked whether that feedback actually improved performance over time. The review concludes that productive feedback requires a supportive culture, stable teacher-student relationships, and deliberate planning of follow-up learning opportunities.

Plain English
Fourth-year medical students learned basic suturing at home using instructional videos, unsupervised practice, and expert feedback on uploaded videos within 72 hours. Final assessment scores (median 24/25) were significantly higher than published benchmarks for in-person expert feedback and video-only learning, and skill retention held at six months. Home-based deliberate practice with asynchronous digital feedback is a viable and effective way to train procedural skills.

Plain English
This paper reviews how AI and large language models are being used in surgical training, covering simulation assessment, virtual and augmented reality platforms, and automated feedback generation. Key opportunities include personalizing feedback and scaling training access across institutions. Remaining challenges include AI accuracy, bias, privacy concerns, and integrating these tools into existing surgical curricula.

Plain English
Physiotherapy students completed a four-session remote training program for upper limb neurodynamic testing, uploading performance videos for instructor feedback after each session. Students reached the target performance standard by session two on checklist measures and session three on rubric scores. Two to three remote sessions are sufficient to develop this clinical skill, making the approach a practical option when in-person training is unavailable.

Plain English
An AI algorithm was trained to assess whether students passed or failed basic laparoscopic simulation exercises by analyzing video footage of two tasks. Agreement with expert evaluators was 79.7% for one task and 93% for the other. These results demonstrate that AI can automate assessment of surgical simulation training at a level approaching expert agreement, addressing the bottleneck of scarce trained evaluators.

Plain English
A remote laparoscopic simulation training program was scaled from one institution to 14 centers across eight Latin American countries, with no on-site instructors required. Between 2019 and 2022, 369 trainees completed the program, generating nearly 6,800 training videos with instructor feedback delivered entirely through a digital platform. This model proves that high-quality procedural training can be distributed at scale using only a few trained remote instructors and a digital platform.

Plain English
Researchers translated and validated the Readiness for Interprofessional Learning Scale into Spanish for use with health sciences students in Chile across six universities. The scale showed good reliability (Cronbach's alpha 0.86) and was sensitive to changes after interprofessional workshops. This gives Spanish-speaking health education programs a validated tool for measuring whether students are prepared to learn alongside colleagues from other disciplines.

Plain English
Ninety-two physiotherapy students worked through eight interactive virtual clinical scenarios to develop musculoskeletal diagnostic reasoning skills. Error rates on pattern recognition dropped significantly from the first to last scenario, with students 2.6 times more likely to correctly identify the clinical pattern by the end. Virtual scenarios allow students to repeatedly practice diagnostic reasoning in a safe environment that traditional case-based teaching does not provide.

Plain English
Biomechanical motion data was collected from seven expert and seven novice anesthesiologists performing simulated central venous access. Experts completed the procedure faster with higher hand speed and acceleration, and these differences varied depending on the procedure stage and which hand was used. These objective biomechanical differences can be used to define specific training targets and track novice progression toward expert-level performance.

Plain English
A medical school gastroenterology course was redesigned between 2008 and 2020 by adding interactive sessions, recorded lectures, and a digital learning platform. Students spent more time engaging with the course and reported significant improvements in feedback quality and access to learning materials. Incremental, technology-supported curriculum changes can meaningfully improve student engagement and satisfaction in theoretical medical courses.

Plain English
A new simulation device for training peripheral venous access was developed using a participatory design approach, involving nurses and educators through iterative cycles of user testing and feedback. The process revealed that involving end users throughout device development—rather than only at the end—produces tools that feel meaningful and are more likely to be adopted. Design and anthropology methods can be applied productively to healthcare simulation equipment development.

Plain English
A paracentesis simulation workshop was run with 247 undergraduate medical students using a purpose-built abdominal model. Students' procedural checklist scores rose from 55.7% before training to 92.9% after, and satisfaction ratings averaged 4.8 out of 5. Simulation-based paracentesis training is effective at rapidly building clinical skill in a safe environment before students encounter real patients.

Plain English
A seven-item perception questionnaire for clinical simulation workshops was adapted and validated with 210 medical students across six simulation programs. Confirmatory factor analysis showed the instrument was unidimensional with good internal consistency (Cronbach's alpha 0.72), and overall student perceptions were strongly positive. This validated tool gives simulation educators a reliable way to measure whether students find their training valuable.