Tak Wing Yu

Public academic notebook

Notes and connections.

Working ideas on physiotherapy education, assessment, teaching and learning, artificial intelligence, virtual reality, simulation, and clinical technology.

Research pathways

Follow the ideas

Explore four connected pathways through the notebook. Each station opens a note; multicolour stations connect more than one theme.

OpenConnected themes

Academic notes

Browse the library

Assessment and simulationObjective Structured Clinical Examination

An OSCE should assess observable clinical competence rather than reproduce a written examination in practical form.

Core principles

  • Align every station with a learning outcome and a specific clinical skill or reasoning process.
  • Standardise examiner prompts, simulated-patient responses, equipment, timing, and room setup.
  • Use a sufficient sample of stations and observations to support a defensible decision.
  • Ensure candidate instructions, examiner guidance, patient scripts, and rubrics are congruent.
  • Pilot stations and train examiners and simulated patients before implementation.

A strong station makes the learner's performance visible while protecting validity, reliability, authenticity, safety, and fairness.

CurriculumFunctional Anatomy Curriculum Development

Functional Anatomy should connect anatomical knowledge with observation and clinical application.

  • Emphasise surface-anatomy palpation, gait, sit-to-stand, squats, heel raises, and step assessment.
  • Help students connect movement quality with possible impairments.
  • Teach consent, appropriate exposure, draping, and respectful palpation explicitly.
  • Use functional movement as a rapid clinical screen alongside focused tests.
  • Balance directly taught content with structured independent learning.
CurriculumMovement Science Development

Movement-science teaching should move from visible function to biomechanical explanation and then to clinical decision-making.

  • Distinguish osteokinematics from arthrokinematics.
  • Explain open- and close-packed positions and joint roll, glide, and spin.
  • Connect muscle synergy, stabilisation, load distribution, and movement efficiency.
  • Ask learners to predict restrictions, compensations, and implications for exercise prescription.
Conference synthesisAMEE 2023 Medical Education Notes

Key themes included microlearning, learner participation, experiential education, student wellbeing, AI, and programmatic assessment.

  • Keep microlearning focused on one objective and avoid overcrowding.
  • Treat learners as stakeholders in curriculum and feedback design.
  • Use community and experiential learning to challenge hospital-centred assumptions.
  • Build belonging, trust, peer support, and workload-management skills into the curriculum.
  • Use multiple assessment data points over time rather than relying only on one high-stakes event.
Teaching and learningStudent Engagement in Health Professions Education

Engagement is influenced by lesson structure, clinical relevance, lecturer presence, feedback, and opportunities for interaction.

  • Break long sessions into purposeful segments and changes of activity.
  • Use clinical cases, brief videos, interactive questions, simulation, and carefully reviewed AI-generated cases.
  • State the topic and intended learning clearly at the start.
  • Use movement, voice, appropriate humour, and responsiveness deliberately.
  • Address barriers such as abstract content, weak interaction, overloaded timetables, and poor attendance.
Learning designTeaching and Learning with Technology

Technology should address a defined educational problem rather than be adopted because it is novel.

  1. Define the learning need and intended outcome.
  2. Select the simplest suitable technology.
  3. Align the activity with assessment.
  4. Protect accessibility, privacy, and fairness.
  5. Pilot, pre-brief, facilitate, debrief, and evaluate.
  6. Refine or discontinue the intervention according to evidence.
Artificial intelligenceAI Foundations and Educational Use

AI includes machine learning, neural networks, natural-language processing, and computer vision. Educational benefits must be considered alongside limitations.

Responsible use

  • Build AI literacy before expecting effective use.
  • Require transparency and appropriate disclosure.
  • Protect personal, sensitive, confidential, and copyrighted information.
  • Review outputs for inaccuracy, fabrication, bias, and missing context.
  • Keep students and researchers accountable for final work.
  • Use AI-content detection cautiously and never as definitive proof by itself.
Generative AIGenerative AI in Teaching and Learning

Generative AI is most educationally useful when it augments human thinking rather than substitutes for it.

  • Throwing: asking for a quick answer with little direction.
  • Flipping: asking AI to question the learner and stimulate thought.
  • Directing: specifying the task, context, standards, and intended outcome.

Learners should plan, iterate, evaluate, disclose material AI use, and remain responsible for accuracy, reasoning, and voice. The first response is material for review, not a finished product.

Research integrityResearch Integrity in the Generative AI Era

AI can function as a preliminary reviewer by identifying unclear arguments, missing context, and possible weaknesses.

  • Disclose material AI contribution.
  • Verify every source, analysis, and generated claim.
  • Do not expose confidential manuscripts or data to unauthorised systems.
  • Treat AI feedback as advisory and challenge it where necessary.
  • Do not delegate novelty, methods, interpretation, or research accountability.
Teaching and learningProductive Struggle and the Safe Bubble

A supportive learning environment should not remove every challenge. The educational task is to distinguish between unnecessary barriers and the kinds of difficulty that help students build judgement, resilience, and professional independence.

  • Ask students to attempt the problem before using AI.
  • Use AI output as material for critique rather than as a finished answer.
  • Make students explain what they accepted, rejected, and changed.
  • Preserve some tasks where memory, reasoning, communication, and practical performance must be demonstrated without AI assistance.
  • Distinguish compassionate support from over-protection.
Virtual realityVirtual Reality Education Design Toolkit

VR should address a genuine learning need and provide value beyond a simpler teaching method.

  1. Identify the educational problem and target learners.
  2. Align the activity with outcomes and assessment.
  3. Define governance and stakeholder responsibilities.
  4. Plan technical, spatial, and operational readiness.
  5. Pre-brief learners and manage cognitive load.
  6. Protect comfort, safety, withdrawal rights, accessibility, equity, and fairness.
  7. Debrief, collect feedback, and refine iteratively.
Clinical technologyPaediatric Exoskeleton: Educational Screening Overview

This public note records educational themes only and is not a clinical protocol or substitute for current manufacturer guidance, clinical assessment, or medical clearance.

  • Confirm device-specific body measurements and fit.
  • Review joint range, alignment, hip status, tone, behaviour, bone health, recent surgery, and weight-bearing restrictions.
  • Confirm that the child can communicate pain or distress, including through non-verbal cues.
  • Escalate uncertain findings to appropriately qualified clinical and medical professionals.
Sport and researchPara-sport Reaction and Cognitive Testing

Reaction testing may examine simple responses, choices between responses, or more complex processing of competing information.

Possible tools include flanker paradigms, Corsi block tasks, the Tower of London, block design, and matrix reasoning. Test selection should follow the research question and account for athlete classification, communication, sensory access, and motor requirements. General cognitive tests should not be treated simplistically as measures of sporting intelligence.

Research digestAI in Physiotherapy Education

Direct evidence about generative AI in physiotherapy education remains limited. The defensible approach is to use AI as a supervised thinking partner while preserving observation, communication, practical examination, and accountable clinical judgement.

Possible teaching applications

  • Ask students to critique an imperfect AI-generated assessment or treatment plan.
  • Require an initial judgement before AI consultation, followed by comparison and reflection.
  • Review and rewrite AI-generated patient education for accuracy, safety, and person-centred communication.
  • Use AI to vary simulation cases within educator-controlled boundaries.
  • Require disclosure of the tool, purpose, verification, and student contribution.

The core sequence is: reason, consult, challenge, decide, and reflect.

AI governanceAI Policy and Academic Integrity Pressure

University AI policy matters, but policy alone is not enough. Students make decisions about AI under pressure from grades, deadlines, peer norms, and online misinformation.

  • State AI permissions clearly for each assessment.
  • Require process, verification, sources, and student contribution where AI use is allowed.
  • Use flawed AI outputs as teaching material.
  • Discuss peer pressure and misinformation directly.
  • Assess reasoning, defence, and process rather than only the final artefact.
Academic workAI and the Acceleration of Academic Work

AI does not simply save time. In academic institutions, efficiency gains are often absorbed into higher output expectations, shifting the real bottleneck toward judgement, direction, and accountability.

  • The key question is not only whether AI was allowed, but whether the user can account for each decision.
  • Expert users can direct AI outputs critically; novice users may bypass the struggle needed to develop expertise.
  • Assessment and supervision should focus more on reasoning, trade-offs, and defensible choices.
  • Reflective AI-use statements should explain judgement, not only prompts.
← Return to the archive