This study explores the use of machine learning to predict neurodevelopmental disorders (NDs) in children using a serious game dataset. The SmartSpeech project, which includes a serious game and a machine learning model, aims to assist clinicians in identifying NDs. The study uses a dataset of 520 instances, reduced to 473 participants after preprocessing. Cluster analysis and reliability analysis were conducted to identify patterns and ensure measurement consistency. A logistic regression model was developed to predict the presence of NDs based on latent factors derived from the data. The model was tested on a subset of 184 participants with an average age of 7 years, achieving high accuracy, precision, recall, and F1-score. The results indicate that the model effectively distinguishes between children with and without NDs. The study highlights the effectiveness of machine learning in diagnosing NDs based on cognitive features and offers new opportunities for decision-making, classification, and clinical assessment. The findings suggest that the model can contribute to early and personalized interventions for at-risk individuals. The study also discusses the advantages and limitations of using machine learning in ND research, emphasizing the need for further studies to improve model accuracy and applicability. The results demonstrate the potential of machine learning in enhancing the accuracy and efficiency of ND diagnosis.This study explores the use of machine learning to predict neurodevelopmental disorders (NDs) in children using a serious game dataset. The SmartSpeech project, which includes a serious game and a machine learning model, aims to assist clinicians in identifying NDs. The study uses a dataset of 520 instances, reduced to 473 participants after preprocessing. Cluster analysis and reliability analysis were conducted to identify patterns and ensure measurement consistency. A logistic regression model was developed to predict the presence of NDs based on latent factors derived from the data. The model was tested on a subset of 184 participants with an average age of 7 years, achieving high accuracy, precision, recall, and F1-score. The results indicate that the model effectively distinguishes between children with and without NDs. The study highlights the effectiveness of machine learning in diagnosing NDs based on cognitive features and offers new opportunities for decision-making, classification, and clinical assessment. The findings suggest that the model can contribute to early and personalized interventions for at-risk individuals. The study also discusses the advantages and limitations of using machine learning in ND research, emphasizing the need for further studies to improve model accuracy and applicability. The results demonstrate the potential of machine learning in enhancing the accuracy and efficiency of ND diagnosis.