Using Artificial Intelligence algorithms and Live GPS to Determine the Fitness Level of Football Players in the Iraqi Premier League

Fitness is one of the essential elements in soccer, and fitness coaches face major challenges, most notably assessing the team's fitness level and understanding and analyzing the huge amounts of GPS data collected from players' performance. Using artificial intelligence algorithms is one of the most important ways to analyze this data accurately and efficiently. In this paper, an improved algorithm (LSTM-XGB) was proposed to predict the fitness level of Al-Talaba Sports Club in the Iraqi Premier League, comparing its performance with three exist algorithms: LSTM, DNN, and GRU, during the two stages of the league. The study relied on real data collected using GPS devices that were attached to players using the vests during the matches. The data were recorded in the memory of the device and then transferred to PlayerTek Plus software through the docking station, and finally exported as CSV files for analysis and model training. The results showed the superiority of the proposed method with an accuracy of (97.6% and 97%), a precision of (97.1% and 94%), and a recall of (96.2% and 95.12%) in the two stages of the Iraqi league, which reflects the effectiveness of the algorithm in assessing and analyzing fitness levels and improving sports performance.