SAVRN
Search Contact SAVRN
Data to Intelligence

Sources and evidence

The research behind
the principle.

Every study and agency document, summarized in plain terms.

Every research figure in this section comes from one of these sources. Each entry says what kind of source it is, what it covers and what it found. The original is one click away, in a new tab.

  1. Systematic review

    Just How Confident Can We Be in Predicting Sports Injuries?

    Bullock et al, Sports Medicine, 2022

    30 studies and 204 injury prediction models in sport, searched to June 2021.

    What it found

    • No study had tested its model on a separate group of athletes.
    • 98% of the models were at high or unclear risk of bias.
    • 3 of the 30 studies worked out their sample size in advance, and the authors could recommend no model for use in practice.

    Cited on: Data to Intelligence, The Method, Athlete Development

    Open original source

  2. Narrative review

    Replication concerns in sports and exercise science

    Mesquida et al, Royal Society Open Science, 2022

    Sample size, statistical power and replication in sports and exercise science research.

    What it found

    • It cites a 2020 audit in which 12 of 120 studies in the Journal of Sports Sciences had worked out their sample size in advance.
    • It reports a median sample size of 19 in the Journal of Sports Sciences and says samples that small are likely underpowered, especially for small and medium effects.
    • It shows that a paired study of 20 people has 45% power to detect an effect size of 0.43, and that 44 people are needed for 80% power.

    Cited on: Data to Intelligence

    Open original source

  3. Replication project

    Estimating the Replicability of Sports and Exercise Science Research

    Murphy et al, Sports Medicine, 2025

    25 studies from top-quartile sports and exercise science journals, published from 2016 to 2021.

    What it found

    • 7 of the 25 studies, or 28%, met all three of the project's tests for a successful replication.
    • Effect sizes were substantially smaller when the studies were repeated.

    Cited on: Data to Intelligence

    Open original source

  4. Methods paper

    Risk factors for sports injuries: a methodological approach

    Bahr and Holme, British Journal of Sports Medicine, 2003

    How to design studies of sports injury risk factors, using hamstring strains as the example.

    What it found

    • Detecting a moderate to strong risk factor takes 20 to 50 injury cases.
    • Detecting a small to moderate risk factor takes about 200 injured athletes.
    • The hamstring studies published to that date were too small to detect small to moderate associations.

    Cited on: Data to Intelligence, The Method

    Open original source

  5. Audit of published research

    Invisible Sportswomen: The Sex Data Gap in Sport and Exercise Science Research

    Cowley et al, Women in Sport and Physical Activity Journal, 2021

    5,261 publications and 12,511,386 participants in six sport and exercise science journals, 2014 to 2020.

    What it found

    • 31% of publications studied men only, 6% studied women only and 63% included both.
    • Women were 34% of all participants, 4,254,445 of 12,511,386.
    • The authors conclude that women remain significantly underrepresented in the research.

    Cited on: Athlete Development

    Open original source

  6. Systematic review

    Current Approaches to the Use of Artificial Intelligence for Injury Risk Assessment and Performance Prediction in Team Sports

    Claudino et al, Sports Medicine Open, 2019

    58 studies of artificial intelligence for injury risk and performance in team sports, covering 12 sports and 11 techniques.

    What it found

    • The pooled sample was 6,456 participants, 97% male and 3% female.
    • 76% of participants were professional athletes.
    • Soccer, basketball, handball and volleyball had the most applications.

    Cited on: Athlete Development

    Open original source

  7. Systematic review

    Rates of Reporting and Analyzing Race and Ethnicity in Athlete-Specific Sports Medicine Research

    Sonnier et al, Orthopaedic Journal of Sports Medicine, 2024

    842 athlete studies in three orthopedic sports medicine journals, 2017 to 2021.

    What it found

    • Age was reported in 84.1% of studies and sex in 87.0%.
    • Race was reported in 3.8% of studies and ethnicity in 2.0%.
    • Of the 13 studies that analyzed race, 8 found significant differences in outcomes.

    Cited on: Athlete Development

    Open original source

  8. Meta-analysis of randomized trials

    Effect of specific exercise-based football injury prevention programmes on the overall injury rate in football

    Thorborg et al, British Journal of Sports Medicine, 2017

    6 cluster-randomized trials of exercise-based injury prevention warm-ups in recreational and subelite soccer.

    What it found

    • Across all 6 trials, the programs lowered the overall injury rate, with a risk ratio of 0.75.
    • The 4 trials of the newer program showed 39% fewer injuries, a rate ratio of 0.61 (95% CI 0.48 to 0.77).
    • The 2 trials of the older program showed no reduction.

    Cited on: Athlete Development

    Open original source

  9. Meta-analysis

    Effects and dose-response relationships of resistance training on physical performance in youth athletes

    Lesinski et al, British Journal of Sports Medicine, 2016

    43 controlled studies of resistance training in athletes aged 6 to 18, published from 1985 to 2015.

    What it found

    • Resistance training had moderate effects on muscle strength and vertical jump.
    • It had small effects on linear sprint, agility and sport-specific performance.
    • The effects differed by sex and by the type of training.

    Cited on: Athlete Development

    Open original source

  10. Systematic review

    Monitoring the athlete training response: subjective self-reported measures trump commonly used objective measures

    Saw et al, British Journal of Sports Medicine, 2016

    56 studies of athletes in training that measured well-being with self-reported and objective measures at the same time.

    What it found

    • Athletes' own reports reflected short-term and long-term training load with better sensitivity and consistency than objective measures.
    • Self-reported and objective measures generally did not agree with each other.

    Cited on: Athlete Development

    Open original source

  11. Systematic review

    Single-Item Self-Report Measures of Team-Sport Athlete Wellbeing and Their Relationship With Training Load

    Duignan et al, Journal of Athletic Training, 2020

    21 studies of single-question well-being measures in adult team sport athletes.

    What it found

    • The questions used most often asked about muscle soreness, fatigue, sleep quality, stress and mood.
    • Their link with training load ranged from none to very large, and was mostly trivial to moderate in the largest studies.
    • The authors ask users to weigh how single questions are applied in athlete monitoring.

    Cited on: Athlete Development

    Open original source

  12. Scoping review

    Associations between growth, maturation and injury in youth athletes engaged in elite pathways

    Parry et al, British Journal of Sports Medicine, 2024

    30 studies of growth, maturation and injury in youth athletes in elite pathways, published since 2000.

    What it found

    • Injury incidence and burden generally rose with maturity, and growth-related injuries peaked during the adolescent growth spurt.
    • Faster growth in height and leg length was linked to more injuries.
    • The authors call for more research in female athletes and in more sports.

    Cited on: Athlete Development

    Open original source

  13. Narrative review

    Growth and Maturation Assessment in Youth Sport: Balancing Benefits and Risks

    Lundberg et al, Scandinavian Journal of Medicine and Science in Sports, 2026

    The benefits and risks of assessing growth and maturation in youth sport.

    What it found

    • Growth assessment can identify athletes in rapid growth, who face more growth-related and overuse injuries.
    • The review finds the risks lie in how results are communicated and in the coaching environment, not in taking the measurements.
    • It proposes five principles: educate those involved, protect privacy, keep participation voluntary, use trained assessors and fit assessment into a child-centered monitoring system.

    Cited on: Athlete Development

    Open original source

  14. Consensus statement

    IOC consensus statement on elite youth athletes competing at the Olympic Games

    Bergeron et al, British Journal of Sports Medicine, 2024

    International Olympic Committee statement on elite youth athletes at the Olympic Games.

    What it found

    • It describes adolescence as a period of non-linear, uneven physical, psychological and social development.
    • It proposes a child-centered model, with guidelines to protect youth athletes' health and well-being in international competition.

    Cited on: Athlete Development

    Open original source

  15. Consensus statement

    International Olympic Committee consensus statement: methods for recording and reporting of epidemiological data on injury and illness in sport 2020

    Bahr et al, British Journal of Sports Medicine, 2020

    International Olympic Committee methods for recording and reporting injury and illness data in sport.

    What it found

    • It sets common definitions and methods so injury and illness data can be compared across studies.
    • It was written after 11 separate consensus statements for single sports or settings had appeared.
    • It came out of a three-day consensus meeting in October 2019.

    Cited on: Athlete Development

    Open original source

  16. Scoping review

    Machine learning approaches to injury risk prediction in sport: a scoping review with evidence synthesis

    Leckey et al, British Journal of Sports Medicine, 2025

    38 studies of machine learning for injury risk prediction in sport.

    What it found

    • Soccer was the most studied sport, and 71% of studies reported the area under the curve.
    • Tree-based methods gave the best predictive performance in 60% of studies, and logistic regression beat machine learning in 4 of 12.
    • Small datasets and inconsistent methods, including how injury was defined, held the models back. Three studies reported scores above 0.9, but their clinical relevance was questionable.

    Cited on: The Method, Athlete Development

    Open original source

  17. Systematic review and meta-analysis

    Comparisons of machine learning models to logistic regression in orthopedic sports medicine are confounded by methodological heterogeneity

    Lu et al, Journal of ISAKOS, 2026

    168 head-to-head comparisons of machine learning and logistic regression from 25 orthopedic sports medicine studies.

    What it found

    • In the 125 comparisons at low risk of bias, machine learning showed no advantage over logistic regression.
    • The authors conclude that no firm recommendation can be made that machine learning is better.

    Cited on: The Method

    Open original source

  18. Cluster randomized trial

    Does load management using the acute:chronic workload ratio prevent health problems?

    Dalen-Lorentsen et al, British Journal of Sports Medicine, 2021

    34 elite youth soccer teams in Norway, girls' and boys', randomized for a 10-month season. Nine teams withdrew soon after, leaving 482 players on 25 teams. The 394 players who answered the health questionnaires were analyzed.

    What it found

    • 18 teams were assigned to plan training with a common workload ratio and 16 to train as normal.
    • Health problems did not fall in the workload group, with a relative risk of 1.01 (95% CI 0.91 to 1.12).
    • It was the first experimental test of managing load by this ratio.

    Cited on: Athlete Development

    Open original source

  19. Systematic review

    Video-based visual feedback to enhance motor learning in physical education: a systematic review

    Modinger et al, German Journal of Exercise and Sport Research, 2022

    11 studies of video feedback in physical education in primary and secondary schools.

    What it found

    • Video feedback appeared to improve motor learning and to work better than verbal feedback alone.
    • The authors say class size, lesson time, equipment, teachers' digital skills and data protection must be considered before using it.

    Cited on: Athlete Development

    Open original source

  20. Validation and field study

    OpenCap: Human movement dynamics from smartphone videos

    Uhlrich et al, PLoS Computational Biology, 2023

    An open system that measures movement from two or more smartphone videos, tested in 10 adults and a field study of 100 people.

    What it found

    • Joint angles were within 4.5 degrees on average of laboratory motion capture.
    • The authors put a lab at more than $150,000 of equipment, against under $700 for the phone setup.
    • Collecting data on 100 people took 8 hours, and processing it took 31 hours of computing.

    Cited on: Athlete Development

    Open original source

  21. Federal agency announcement

    FTC Finalizes Changes to Children's Privacy Rule Limiting Companies' Ability to Monetize Kids' Data

    Federal Trade Commission, January 16, 2025

    Amendments to the Children's Online Privacy Protection Rule, announced January 16, 2025.

    What it found

    • Covered online services need verifiable parental consent before collecting personal information from children under 13.
    • They need separate parental consent before disclosing a child's data to third parties, such as for targeted advertising.
    • Data may be kept only as long as reasonably necessary, and personal information now includes biometric identifiers.

    Cited on: Athlete Development

    Open original source

  22. Federal guidance

    Joint Guidance on the Application of FERPA and HIPAA to Student Health Records

    US Department of Health and Human Services and US Department of Education, 2019

    Joint guidance from the US Departments of Health and Human Services and Education, December 2019 update.

    What it found

    • At most colleges, the health records a campus clinic keeps on students are education or treatment records under FERPA.
    • Those records are excluded from HIPAA, even when the school is a HIPAA covered entity.

    Cited on: Athlete Development

    Open original source

Continue

Back to
the section.

The principle sets out what a useful record holds. The method and the design case show how the evidence above shapes it.

The principleHow data becomes intelligence The methodFrom a dataset to a reliable model Design caseAthlete development