Lighthouse AdvisorySLED AI Adoption Intelligence

Higher Education & ResearchIssue 275 records

College Athletics

Scrutiny and measurement controls dominate today’s college-athletics AI signal. Childs et al.’s SportsHCI contextual-integrity study (15 NCAA DI athletes) maps seven continuous-wearable information flows and flags athlete discomfort with automatic sleep/recovery sharing even as workout auto-uploads to coaches are normalized—operationalizing consent after recent wearable-governance coverage. On the performance stack, a Scientific Reports analysis of 32,952 NCAA DI CMJ trials shows processing-pipeline choice alone pushes jump height beyond a 2 cm smallest worthwhile change in about one-third of trials, recommending locked variance-minimization body-weight and derivative onset before any CMJ-derived AI. Frontiers work on 170 Malaysian university athletes finds a relative-age effect but null Holomotion AI loading-risk differences by birth quartile—warning against treating markerless risk scores as selection explanations. Operationally, UT Arlington’s Sep 28 FanWord Assist deal extends mid-major content AI without public quality KPIs. Venable’s July 2026 NIL guidance adds contracting controls for AI training on likeness, voice, biometrics, and synthetic media. Gaps: measured U.S. FanWord edit/error rates; criterion-validated CMJ pipelines; NCAA prospective tests of markerless loading AI.

Evidence records
5
Cross-source patterns
2
Evidence classes
3 academic research1 vendor claim1 standards or public-body guidance
Outcomes
3 cautionary1 emerging1 mixed
Source freshness
3 recent1 older, newly relevant1 new this fortnight
Research completed
2026-10-05

Choose a role to see its takeaway beside every record in the ledger.

Patterns across the evidence

Synthesis · Lighthouse Advisory interpretation

2 patterns, each supported by at least two sources

  1. Lock measurement and validate AI risk labels before using them for athletics decisions

    NCAA DI CMJ pipeline variation can exceed smallest worthwhile change within a trial, and Holomotion AI loading-risk indices failed to track relative-age selection asymmetry in university athletes—so departments should pin processing versions and forbid treating unvalidated AI risk scores as clinical or talent explanations.

    Operating questionIs every AI-ready athletic metric tied to a pinned processing/label definition that has been tested against known biases and SWC thresholds?

    Supporting evidenceAuburn University Athletics collaborators / Scientific Reports (Nature)Sultan Idris Education University / National Sports Institute of Malaysia / Frontiers in Sports and Active Living

Evidence ledger

Full record · every source keeps its link and limitations

5 records
  1. Academic researchCautionaryNewly relevant · Nov 2025

    SportsHCI study maps seven continuous-wearable information flows in NCAA DI and flags intrusive sleep/recovery auto-share

    Childs, Brewer, Thomas et al. analyzed 15 semi-structured interviews with elite U.S. collegiate student-athletes using contextual integrity and identified seven information flows for continuous wearable data among coaches, athletes, teammates/peers, family, and others. Athletes described need-to-know and coach-requested sharing of physiological metrics, automatic post-workout uploads via platforms such as Garmin Clipboard, and stronger discomfort with automatic sleep/recovery sharing (e.g., WHOOP). Authors recommend aligning privacy protections with sport-tech adoption; limitations include organic (not wearable-focused) prompts and a single DI context with women’s-sport skew.

    University of Florida / SportsHCI 2025 (ACM)United States (NCAA Division I interview sample)November 17–19, 2025 (SportsHCI 2025 proceedings)

    Why it matters, evidence and limitations
    Why it matters
    Why now: After 2026-10-01/10-02 wearable-injury ML caution and NCAA performance-tech governance coverage, this previously unarchived SportsHCI paper supplies athlete-side sharing norms that operationalize consent, purpose limits, and selective mute—directly informing DI wearable AI deployments.
    Evidence and measured results
    Opened author PDF reports methods (15 interviews; contextual integrity coding), seven flows (coach/athlete/network receivers), automatic workout upload (TP4) vs intrusive recovery/sleep concerns, limitations section, and NSF/UF AI-Powered Athletics funding acknowledgments. DOI 10.1145/3749385.3749387.
    Limitations and uncertainty
    Qualitative n=15; single DI program ecology; wearable mentions emerged organically; not a security penetration test; preprint-marked author PDF mirrored to ACM proceedings.
    Peer-reviewed qualitative conference paper (author PDF) Role takeaways & implementation
  2. Academic researchCautionaryRecent

    Scientific Reports: CMJ processing pipeline choice exceeds 2 cm SWC in ~1/3 of 32,952 NCAA DI trials

    Goodlett, Roper, Aakur et al. applied 30 factorial CMJ pipelines (5 body-weight × 6 onset methods) to 32,952 countermovement-jump trials from 579 NCAA Division I athletes across 15 teams, 12 sports, and five seasons (11,410 sessions). Variance-minimization body-weight methods were interchangeable (95% LoA within ±5 N). First-derivative onset was most consistent across signal-quality tiers; threshold onsets showed quality-dependent timing bias. Pipeline choice produced jump-height divergence exceeding a 2 cm smallest worthwhile change in about one-third of trials, while matched-pair between-method bias on plausible trials stayed ≤0.4 cm. Authors recommend variance-minimization BW with derivative onset for arms-akimbo field CMJ on consistency—not criterion-validated accuracy—grounds.

    Auburn University Athletics collaborators / Scientific Reports (Nature)United States (NCAA Division I; Auburn Athletics testing dataset)August 19, 2026

    Why it matters, evidence and limitations
    Why it matters
    Why now: Same-week stream vigilance on wearable/injury ML metrics makes this unarchived Aug 2026 DI-scale finding newly actionable—measurement pipeline drift can dominate claimed AI readiness signals before models are even trained.
    Evidence and measured results
    Opened Nature HTML reports abstract metrics (32,952 trials; 579 athletes; 30 pipelines; 2 cm SWC exceedance ~1/3; ≤0.4 cm matched-pair bias), publication date 2026-08-19, Auburn Athletics acknowledgments. PDF fetch returned non-PDF challenge page; HTML used as primary inspection.
    Limitations and uncertainty
    Consistency recommendation without criterion force-plate gold standard; arms-akimbo CMJ only; single athletics department data source; HTML inspection (PDF binary blocked).
    Peer-reviewed large-scale measurement study (HTML) Role takeaways & implementation
  3. Vendor claimEmergingNew this fortnight

    UT Arlington Athletics adopts FanWord Assist for AI-supported Maverick storytelling

    On 2026-09-28 UT Arlington Athletics announced a FanWord Assist partnership to expand digital storytelling for student-athletes, coaches, alumni, and supporters while increasing communications efficiency. Athletics director Jon Fagg and FanWord CEO Christopher Aumueller emphasize authentic storytelling and staff efficiency. The release states FanWord partners with more than 250 NCAA institutions and conferences; About FanWord text on the same page also cites more than 200 athletics organizations. No measured hours saved, edit rates, or quality KPIs are published.

    UT Arlington Athletics / FanWordUnited States (Texas; WAC DI)September 28, 2026

    Why it matters, evidence and limitations
    Why it matters
    Fresh Sep 28 DI FanWord Assist URL not in archive—extends the post-FDU/East Texas communications-AI wave with another mid-major adoption still lacking public outcome metrics.
    Evidence and measured results
    Opened athletics news page (WebFetch + curl HTML) includes dated byline 9/28/2026, Fagg/Aumueller quotes, FanWord Assist scope (features, spotlights, alumni/donor stories), and vendor network claims (250+ / 200+).
    Limitations and uncertainty
    Department PR with vendor quotes; conflicting 250 vs 200 network figures on same page; no independent productivity study.
    University athletics department announcement Role takeaways & implementation
  4. Standards or public-body guidanceCautionaryRecent

    Venable: NIL deals now require AI input/output, synthetic media, and enterprise vendor controls

    Armand J. Zottola’s 2026-07-20 Venable Quick Byte argues NIL has become a commercial ecosystem spanning sponsorships, IP, data monetization, AI, and tech deals. It flags implied association via school marks, unresolved group-licensing questions, and AI vendor demands for athlete images, voice, biometrics, performance analytics, and archives to train systems and create synthetic media. Older consents often omit synthetic voice/avatars/AI endorsements. Author recommends enterprise-style vendor review (security, SLAs, termination), financial diligence on early-stage NIL tech firms, and principle-based governance (centralized review, cross-department committees, escalation) amid unsettled NCAA/state rules.

    Venable LLPUnited States (higher-education athletics counsel audience)July 20, 2026

    Why it matters, evidence and limitations
    Why it matters
    Newly relevant against archived NACUA NIL Go antitrust analysis: shifts from algorithmic FMV clearinghouses to contracting controls for generative/synthetic athlete media and biometric training rights still absent from many athletic department playbooks.
    Evidence and measured results
    Opened Venable HTML includes July 20, 2026 byline, sections on institutional IP, group licensing, AI/synthetic media, vendor procurement, and governance checklist items (centralized review, committees, workflows, disclosure, training, escalation).
    Limitations and uncertainty
    Law-firm advisory, not empirical evaluation; no measured university outcomes; regulatory landscape described as unsettled.
    Law-firm practice guidance (IP Quick Bytes) Role takeaways & implementation
  5. Academic researchMixedRecent

    Frontiers: Malaysian university RAE confirmed, but Holomotion AI loading-risk indices show no birth-quartile differences

    Moganan et al. assessed 170 national-level Malaysian university athletes (116 male, 54 female; mean age 18.6) with Holomotion markerless motion capture, deriving total movement score, exercise risk score, joint-pain risk index, and ligament-strain risk index (modelled tissue loading, not clinical injuries). Significant relative age effect overall (χ²(3)=11.74, p=.008, w=0.26) and in males (p=.036; Q1:Q4=2.06), not females (p=.286; underpowered). Kruskal–Wallis tests found no birth-quartile differences on any AI-derived loading outcome (all p≥.720; η²≈0); binary high-risk classifications similarly null. Authors caution cross-sectional design cannot speak to prospective injury.

    Sultan Idris Education University / National Sports Institute of Malaysia / Frontiers in Sports and Active LivingMalaysia (national-level university athletes); international open-access journalJuly 22, 2026

    Why it matters, evidence and limitations
    Why it matters
    Unarchived July 2026 university-athlete AI screening result that contradicts the intuition that selection asymmetries leave AI-detectable loading signatures—useful control evidence for athletics buyers of markerless risk AI after recent high-metric injury-model editions.
    Evidence and measured results
    Opened Frontiers PDF states methods, χ²/RAE statistics, Kruskal–Wallis nulls across four AI outcomes, modelled-vs-clinical caveat, and limitations (cross-sectional; female power; Malaysian context).
    Limitations and uncertainty
    Malaysia university/national pathway—not NCAA; cross-sectional; AI indices are modelled loading; female RAE non-significance likely Type II; transferability limits apply.
    Peer-reviewed cross-sectional study (PDF) Role takeaways & implementation

How to read this edition

Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.

Academic research
Research produced through an academic institution or peer-reviewed venue.
Vendor claim
A supplier-provided assertion that has not been upgraded to independent evidence.
Standards or public-body guidance
Normative or advisory guidance from a standards body or public institution.