Using Artificial Neural Networks (ANN) to predict decision-making under psychological pressure among Iraqi premier league goalkeepers

Introduction and Problem Statement: The study addresses the shift in modern sports towards digital technology and big data. It highlights a critical gap in traditional performance analysis, which typically relies on "post-event" evaluation and overlooks the mental state preceding a goalkeeper's error. The research posits that decision-making errors often result from "cognitive blindness" caused by psychological pressure, a complex interaction that traditional statistics fail to predict. Objectives: The primary objective was to construct an Artificial Neural Network (ANN) model capable of predicting the accuracy of goalkeepers' decisions based on psychological pressure variables (cognitive and somatic anxiety). The study also aimed to identify the relative importance of these variables in influencing performance. Methodology Approach: The researcher used a descriptive predictive approach. Sample: The sample consisted of 24 main goalkeepers from the Iraqi Premier League for the 2025-2026 season. Tools: The study utilized the Competitive State Anxiety Inventory-2 (CSAI-2) to measure anxiety levels and video analysis to evaluate 480 specific decisions. Modeling: An ANN was built using Python and SPSS Modeler, with inputs including anxiety levels, match importance, timing, and score difference. Key Results High Anxiety: The participants showed high levels of cognitive anxiety (mean 24.5), indicating significant mental load. Model Accuracy: The ANN model achieved a predictive accuracy of 84.0% in the testing phase, demonstrating its capability to handle non-linear data effectively. Key Influencers: "Cognitive Anxiety" was identified as the most significant factor causing errors (38% relative importance), followed by match timing (25%). Conclusions: The research concludes that performance collapse under pressure is not random but can be mathematically modeled. The ANN model proved to be more accurate and objective than traditional human evaluation, which is often subject to bias. The study recommends integrating AI-based analysis into training programs to better manage cognitive load.