Scottish Medical Journal

Accurate triage of lumbar spine magnetic resonance imaging (MRI) referrals for sciatica is important for patient assessment, diagnosis and surgical planning. This study evaluates the accuracy and speed of large language models (LLMs) in automatically vetting lumbar spine MRI referrals from general practice.

Three LLMs (GPT-4, Claude Opus, Gemini) were tasked with assigning an outcome (Accept – Routine, Accept – Urgent, Reject) and flagging MRI contraindications for lumbar spine referrals. Three prompts of increasing detail, including clinical guidelines and training examples, were used. Two radiology registrars synthesised 120 referrals, vetted by two board-certified radiologists, with a third resolving disagreements. Performance was assessed using accuracy, precision, recall and F1 scores.

Inter-rater agreement between radiologists was substantial for vetting outcome (Cohen's κ = 0.76) and contraindication detection (κ = 0.68). Claude Opus with the full prompt achieved the highest accuracy (0.86) for vetting outcomes. GPT-4 with the instruction-only prompt achieved the highest F1 score (0.88) for contraindication detection. LLMs completed the task substantially faster than radiologists (9.8 ± 1.0 vs 135.0 ± 45.0 min).

LLMs demonstrate promising performance in vetting radiological referrals for sciatica, particularly with detailed context. All models identified all urgent referrals, suggesting potential for prioritising vetting worklists and improving timeliness of care.

Effective communication and shared decision-making are essential for optimising urological care, making informed decisions, and improving patient outcomes. The integration of artificial intelligence (AI) in urology has the potential to act as a supportive tool in this process. This review aims to evaluate how AI-based tools support and enhance patient-provider communication and shared decision-making within urological care.

Following PRISMA 2020 guidelines, a systematic search was performed using Cochrane, EMBASE, MEDLINE, and Scopus for literature published between 2019 and 2024. Search terms included ‘Artificial Intelligence’, ‘Urology’, ‘Shared Decision-Making’, and ‘Communication’. Studies were screened using our predefined inclusion and exclusion criteria. Three primary themes were identified, through which the studies were analysed.

Of 807 identified studies, 14 were appropriate for inclusion. Only 14 studies met criteria because most excluded articles did not evaluate AI tools designed for communication, health literacy, or shared decision-making. AI-driven tools, particularly large language models (LLMs), show the potential to reduce knowledge gaps for diverse literacy levels and improve patient comprehension. These aids may improve the readability of complex medical content and translate information with cultural sensitivity. AI may also enhance communication between patients and healthcare providers by automating repetitive tasks, such as responding to frequently asked questions. However, AI has limitations, with different LLMs displaying variable levels of effectiveness and accuracy across urological conditions.

The integration of AI has the potential to enhance communication and promote shared decision-making in urology. However, patients should use AI as a complement to physicians rather than a replacement. To confidently determine their role and ensure AI output accuracy, further studies, including validation against clinical standards and real-world accuracy are required.

To deliver and evaluate two Emergency Urology Skills Training (EUST) courses in Ethiopia, aimed at equipping surgical and urology residents with hands-on skills and confidence in managing urological emergencies in resource-limited settings.

Two one-day, practical training courses were held in Hawassa and Addis Ababa in November 2024 and May 2025. Pre-course questionnaires assessed delegates’ baseline confidence, prior training and the utility of a pre-course manual. A blended curriculum comprising didactic lectures, skill stations and one-to-one mentorship was delivered by a collaborative team of local and international faculty. Post-course evaluations measured improvements in knowledge, confidence and satisfaction.

Twenty-three participants from each centre completed matched pre- and post-course multiple-choice questionnaires assessing knowledge of emergency urology procedures. Pre-course exposure to structured skills training was limited (≤30%). Both groups showed statistically significant improvements in post-course scores (Hawassa: p = 0.002; Addis Ababa: p = 0.007). Self-rated confidence and knowledge improved significantly (p < 0.05) in six of eight core procedures.

The EUST model effectively improved trainees’ confidence and procedural competence in emergency urology. With adequate support, this model is scalable and applicable to other resource-limited countries seeking to strengthen urological emergency care and their training capacity.

Transient isolated right ventricular hypertrophy (RVH) is an infrequently recognized cause of early neonatal cyanosis and prolonged oxygen requirement, closely mimicking cyanotic congenital heart disease or persistent pulmonary hypertension of the newborn.

We present a narrative review of published neonatal cases of transient isolated RVH and describe illustrative observations from four term neonates managed at a tertiary neonatal unit over a 3-year period. Clinical features, antenatal exposures, echocardiographic findings, management, and outcomes were reviewed. A focused literature search identified previously reported neonatal cases.

Including the present cohort, 18 neonatal cases have been described to date. Most neonates presented within the first day of life with cyanosis. Echocardiography demonstrated isolated RVH with preserved biventricular function and absence of structural heart disease; premature ductal closure was documented in several cases. Identified risk factors included fetal or perinatal distress, maternal exposure to nonsteroidal anti-inflammatory drugs or corticosteroids, maternal diabetes, and polyphenol-rich substances. Clinical improvement occurred with supportive care, and follow-up echocardiography showed regression of RVH within 4–12 weeks.

Transient isolated RVH represents a benign, self-limiting cause of early neonatal cyanosis. Awareness of this phenotype and its natural history may help avoid unnecessary investigations/interventions.