{"resourceId":"cao-posesync-taichi-ai-rct-2026-08-29","versions":[{"version":"external-38093a9f0f4a15ab4bdb63e7ac63d2246fe1d70a9fe431192a6ff8ef383480a8","resource":{"id":"cao-posesync-taichi-ai-rct-2026-08-29","title":"University PE RCT: PoseSync Vitality pose-estimation Tai Chi tutor beats matched video instruction on expert-rated skill and retention","organization":"Universiti Sains Malaysia / Education and Information Technologies (Springer)","sector":"University physical education / AI movement tutoring","geography":"Malaysia (university PE course); international journal","publishedAt":"August 29, 2026","publicationDate":"2026-08-29","eventDate":null,"sourceName":"Education and Information Technologies","sourceLabel":"Peer-reviewed randomized controlled trial (abstract)","sourceUrl":"https://link.springer.com/article/10.1007/s10639-026-14132-8","evidenceClass":"academic-research","outcomeClass":"mixed","topics":["developers-agents","accessibility-workforce","operating-model","knowledge-work"],"finding":"Cao, Chau, Wan Yahaya et al. randomly assigned 120 undergraduates in a compulsory university Tai Chi PE course to PoseSync Vitality (AI pose estimation with joint-specific cues and practice summaries) or dose-matched video instruction. The AI group showed larger expert-rated performance gains from pretest to posttest and to a two-week retention test, higher Learner Interface and Personalization scores, and higher posttest Performance Consistency Score (PCS) versus video. Full numeric effect sizes are not visible on the opened abstract page; ethics approval USM No. 25050453 is stated.","sledRelevance":"Newly archived RCT of AI movement tutoring in university PE—complements the stream’s Frontiers PE LLM+RAG quasi-experiment and answers Pass A demand for measured instructional AI benefits with a control condition, while remaining outside varsity medical AI.","evidence":"Springer abstract states n=120 RCT, PSV vs video, expert ratings at pretest/posttest/retention, satisfaction scales, and PCS biomechanical alignment logs; results sentence confirms larger expert-rated gains and higher personalization/PCS for PSV. Opened page is preview/abstract; figures listed but body paywalled.","architectureImplications":"Requires camera/pose-estimation pipeline, expert reference trajectories, real-time corrective UX, and logging of PCS—deployable in PE classrooms more readily than full athlete databanks, but needs privacy controls for continuous video.","governanceImplications":"Treat as instructional assist under PE faculty ownership; obtain student consent for video capture; do not conflate course skill gains with medical clearance or roster selection.","securityPrivacyImplications":"Pose video and biomechanical traces are biometric-adjacent; minimize retention, restrict coach/staff access, and ban secondary marketing reuse.","caveats":"Abstract-only metrics on opened page; Malaysian compulsory PE not NCAA athletics; Tai Chi domain may not transfer to contact sports; no injury/outcome endpoints.","streamIds":["college-athletics"],"roles":{"sales":"Sales framing: University PE and developmental athletics units want individualized movement feedback without adding instructors. Lead with a classroom pilot comparing AI pose tutoring to existing video modules on one skill block, using expert ratings and retention—not injury reduction—as success. Bound claims to instructional gains shown in this RCT setting. Unsupported: guaranteed varsity performance lift, medical decision support, or NCAA-wide transfer from a Tai Chi PE course.","engineering":"Engineering framing: Fit is real-time pose estimation with joint cues and practice summaries in a PE LMS/classroom stack. Architecture needs calibrated cameras, expert trajectory library, latency budget for live feedback, and opt-in video storage. Prerequisites: course learning objectives and expert rating rubric. Constraints: lighting/occlusion, paywalled method details, and non-NCAA population. Security: encrypt video, short retention, role-based review. Proof of value: expert-rated delta vs video control plus student opt-out rate.","delivery":"Delivery framing: Pilot one PE section with matched curriculum dose, dual expert raters, and retention retest. Dependencies: instructor buy-in, AV kit, IRB/privacy review for classroom video. Skills adoption trains instructors to override bad cues. Governance checkpoints: consent script, weekly false-cue review, stop rule if injury complaints rise. Risks: overclaiming medical utility and biometric retention creep."},"roleDetails":{"sales":{"customerProblem":"Large PE/developmental cohorts cannot get individualized movement feedback from scarce instructors.","stakeholders":["PE faculty","Athletics performance staff (developmental)","Registrar/academic affairs","Privacy officer"],"discoveryQuestions":["Which skill block has a stable expert rubric?","Where will video be stored and for how long?","What retention test proves durable learning?"],"valueHypothesis":"Pose-estimation tutoring may improve expert-rated skill and short-term retention versus video alone if instructors remain responsible for grading and safety.","potentialEngagement":"One-term PE section pilot with matched video control and dual expert ratings.","unsupportedClaims":["Varsity win/injury outcomes","NCAA medical clearance use","Guaranteed transfer beyond Tai Chi PE"]},"engineering":{"fit":"Classroom AI movement tutor with pose estimation—not clinical wearables.","architectureIntegration":"Camera capture → pose model → corrective UX → PCS logs integrated with course LMS.","prerequisites":["Expert reference library","Rubric","Consent workflow","AV hardware"],"constraints":["Occlusion/lighting","Abstract-limited method detail","Non-U.S. study population"],"security":"Minimize biometric video retention; encrypt at rest; ban training reuse without consent.","proofOfValue":["Expert-rated gain vs video","Retention retest delta","Opt-out/complaint rate"]},"delivery":{"work":"Stand up camera kit, train instructors on override, run RCT-like pilot, publish internal learning report.","dependencies":["Faculty owner","Privacy review","AV support","Rater training"],"ownership":"PE course instructor owns instructional decisions; IT owns video platform.","skillsAdoption":"Teach staff to treat cues as assistive and to re-demonstrate when PCS conflicts with expert eye.","governanceCheckpoints":["Consent approval","Midterm cue-error review","End-term retention gate"],"acceptanceCriteria":["Completed retention test","Documented expert inter-rater process","Written video retention limit"],"risks":["Privacy complaints","Overgeneralization to varsity medicine","Deskilling of observation teaching"]}},"retrievedAt":"2026-10-02T22:18:42-05:00","enrichedAt":"2026-10-02T22:20:30-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"May expand individualized feedback in large PE classes while requiring new instructor AI-literacy and risking reduced observational teaching practice.","procurementImplications":"Buy against classroom learning metrics and privacy schedule; do not purchase under sports-medicine injury RFPs.","operatingModelImplications":"Shifts PE from one-to-many demo toward AI-augmented practice with instructor override as the operating norm.","sourceVerification":{"responseId":"curl-posesync-20261002T221842","openedUrl":"https://link.springer.com/article/10.1007/s10639-026-14132-8","referenceExcerpt":"the PSV group had larger expert rated performance gains from pretest to posttest","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}