🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Grading 1800 student assignments costs ~$31K in faculty time. Pairing ChatGPT with a teaching assistant drops that to just $1.7K. That’s a $29K annual saving for one cohort. LLMs are more than hype if you know how to use it strategically. #MedEd

Jane tried ChatGPT for a sore knee. It felt helpful; quick, easy, useful. Weeks later, she felt emotionally low and typed a message… then paused. “Do I really want to share this?” AI in healthcare isn’t just about utility. It’s about trust.

Most assessments test facts. Few test #clinicalreasoning in uncertainty. Enter: Script Concordance Test. Designed for decision-making when the path isn't clear. Now, AI (ChatGPT-4o, Claude) can generate SCTs with 90% expert approval. #MedEd just got way more scalable.

Some students feel guilty using ChatGPT. But that guilt isn’t always harmful, it can be useful. It nudges them to ask: “Am I using AI to think better—or to avoid thinking?” Teach this: Use GenAI to augment, not replace your mind.

Myth: Retrieval-augmented generation (RAG) always boosts LLM accuracy. In USMLE-style pharmacology MCQ writing, GPT-4o without RAG (feeding in external sources) scored 88% accuracy, higher than RAG’s 69.2%. Carefully built prompts > retrieval hacks.

Expert-designed interprofessional education scenarios took 2 hours each. GPT-4o, using a 3-step refinement loop, produced higher-quality cases in 9 minutes. Not by guessing, but by simulating students, integrating feedback, and iterating like faculty teams. #MedEd

Supervisors scored identical diagnostic decisions 1 full point lower when the clinical outcome was bad. Same reasoning, same choice; just worse result. Outcome bias is real. And it shapes the feedback learners receive. Focus on reasoning process, not result.

Clinician-educators who ignore AI are falling behind. Engage now, or risk training a generation that relies on tools they don’t understand. It is crucial to avoid becoming mere “humans-in-the-loop”. Otherwise, we risk losing the core of medicine: The care of the patient.

Students could use AI for feedback. Many didn’t. In a pharmacy course: * 67% used AI in round 1 * 56% in round 2 Peer feedback was rated more relevant, actionable, and specific. Insight: Without trust and context, students won’t use AI, even when it’s allowed.

Researchers gave ChatGPT-4o more info pulled from textbooks. Its accuracy performance got worse. Read the latest post in my free newsletter: medicaleducationflamingo.substack.com/p/ai-performed…

211 experts from 79 countries built a global digital health framework for #MedEd. It defines 19 competencies across 4 domains: 33 mandatory, 145 discretionary learning outcomes.

This is what I see from my office window today.

Medical ed won’t be transformed by AI tools alone. Without: * Institutional AI teams with authority * Funding models beyond “bake sale budgets” * Rigorous multi-site research We risk repeating the EHR debacle: high hopes, low impact.

Calling all researchers: if you're using AI for peer review, please read this thread. #MedEd

Compared to many humans' “facts”, current LLMs like o3 are basically professors.

Spanish Journal of Medical Education is an ESCI-indexed, open-access journal with no article processing charges. Submission-to-publication times are shorter than those of many #MedEd journals. It is gaining increasing attention. Consider sending your papers.

AI outperformed human experts in bronchoscopy training. 30 min with an AR/AI simulator made ICU docs navigate lungs 23% faster & 7 fewer wrong turns than expert-led peers. The power of structured, instant AI feedback for a psychomotor skill training. #MedEd

Our students' study has been published in Anatomical Sciences Education today! Want to lower stress in anatomy class? Let students create the content. A 3D modeling elective let our medical students build their own anatomical models: VR, 3D printing, animation.

Web of Science (Clarivate) just released the latest (2024) journal impact factors. #MedEd journals; Academic Medicine, Medical Education, and Medical Teacher have very similar impact factors this year.

Dental students coached by ChatGPT outperformed video-only peers in VR drills (73 vs 66, +12%) with 56% lower cognitive load (measured by eye tracking). Time to experiment AI coaches worldwide. #MedEd