A causal inference method for athletic injuries based on quantile threshold functions and latent Gaussian DAG models

Area of Science:

• Sports Medicine
• Biostatistics
• Epidemiology

Background:

• Causal inference is crucial for developing effective athletic injury prevention strategies.
• Directed acyclic graph (DAG) models are increasingly used to study athletic injuries.

Purpose of the Study:

• To propose and validate a quantile threshold function (QTF) integrated with a latent DAG model for ordinal variables.
• To estimate ordinal causal effects (OCE) related to athletic injuries.

Main Methods:

• Transformed continuous variables into ordinal variables to construct a DAG.
• Applied a latent causal inference framework to analyze the DAG and estimate OCE.
• Utilized DAG path analysis to identify direct and indirect injury pathways.

Main Results:

• The proposed method demonstrated significant group differences (F > 52,000, P < 0.05) on real-world data.
• Identified three direct and two indirect causal pathways contributing to athletic injuries.
• Quantified the impact of causal pathways on injury risk through intervention.

Conclusions:

• The novel QTF-integrated latent DAG approach provides significant theoretical and methodological insights into athletic injuries.
• This framework is essential for optimizing training programs and mitigating injury risk.
• The study establishes a robust basis for future research in sports injury causal inference.