A hybrid deep learning framework combining CTGAN and transformers for personalized injury mitigation in elite sports.
- 1School of Physical Education, Harbin Normal University, Harbin, Heilongjiang, China.
- 2Education College, Wenzhou Business College, Wenzhou, China.
Abstract
Introduction
Optimizing training in elite sports while managing injury risk requires models that can learn from heterogeneous and imbalanced athlete records without contaminating evaluation data. This study presents PTI-GANFormer (Personalized Training and Injury Risk Prediction using Generative Adversarial Network-Former), an integrated decision-support framework. Its contribution is not CTGAN or self-attention in isolation; rather, it is the leakage-controlled combination of training-only conditional augmentation, three functionally separated attention pathways, five domain-engineered indicators, and a transparent risk-to-action recommendation layer.Methods
After median imputation, plausibility-aware outlier handling, categorical embedding, scaling, and feature engineering, the data were stratified into training and untouched testing partitions. CTGAN was fitted only to the training partition and generated minority-class records until each training class matched the original majority count of 7,894. The hybrid classifier distinguishes No Injury, Minor Injury, and Major Injury, while the recommendation engine converts predicted risk into advisory workload and recovery options that require practitioner review.Results
Across repeated stratified evaluation runs, PTI-GANFormer achieved an accuracy of 0.960 ± 0.004 (95% CI: 0.957-0.963), precision of 0.950 ± 0.006, recall of 0.960 ± 0.005, F1-score of 0.950 ± 0.005, and macro ROC-AUC of 0.981 ± 0.004.Discussion
These results support predictive risk stratification and decision support; they do not by themselves establish injury prevention or performance enhancement.
Related
- DeepOSSA: a deep learning framework for personalized survival prediction in osteosarcoma patients
- MuscleSCS: a standardized programming framework combining neural neuromodulation and spinally mediated paraspinal muscle modulation
- Athlete-AI: a multi-modal artificial intelligence framework for personalized athlete monitoring and injury risk forecasting
- COMBINING CHEST X-RAY AND CLINICAL COVARIATES FOR OSTEOPOROSIS SCREENING: A MULTIMODAL DEEP LEARNING APPROACH
- A deep learning-based model combining quantitative parameters of initial chest computed tomography and clinical variables for the prediction of posttraumatic acute respiratory distress syndrome
BibTeX
@article{teng2026hybrid,
title = {{A hybrid deep learning framework combining CTGAN and transformers for personalized injury mitigation in elite sports.}},
author = {Tiemin Teng and Lei Zhang},
journal = {Frontiers in physiology},
year = {2026},
month = oct,
doi = {10.3389/fphys.2026.1923556},
url = {https://europepmc.org/article/MED/42908835},
note = {CC BY},
}
RIS
TY - JOUR
TI - A hybrid deep learning framework combining CTGAN and transformers for personalized injury mitigation in elite sports.
AU - Tiemin Teng
AU - Lei Zhang
PY - 2026
DA - 2026/10/07
JO - Frontiers in physiology
DO - 10.3389/fphys.2026.1923556
UR - https://europepmc.org/article/MED/42908835
AB - Introduction Optimizing training in elite sports while managing injury risk requires models that can learn from heterogeneous and imbalanced athlete records without contaminating evaluation data. This study presents PTI-GANFormer (Personalized Training and Injury Risk Prediction using Generative Adversarial Network-Former), an integrated decision-support framework. Its contribution is not CTGAN or self-attention in isolation; rather, it is the leakage-controlled combination of training-only conditional augmentation, three functionally separated attention pathways, five domain-engineered indicators, and a transparent risk-to-action recommendation layer.Methods After median imputation, plausibility-aware outlier handling, categorical embedding, scaling, and feature engineering, the data were stratified into training and untouched testing partitions. CTGAN was fitted only to the training partition and generated minority-class records until each training class matched the original majority count of 7,894. The hybrid classifier distinguishes No Injury, Minor Injury, and Major Injury, while the recommendation engine converts predicted risk into advisory workload and recovery options that require practitioner review.Results Across repeated stratified evaluation runs, PTI-GANFormer achieved an accuracy of 0.960 ± 0.004 (95% CI: 0.957-0.963), precision of 0.950 ± 0.006, recall of 0.960 ± 0.005, F1-score of 0.950 ± 0.005, and macro ROC-AUC of 0.981 ± 0.004.Discussion These results support predictive risk stratification and decision support; they do not by themselves establish injury prevention or performance enhancement.
SN - 1664-042X
ER -