توسعه یک مدل هوش مصنوعی مبتنی بر درخت تصمیم برای افزایش تشخیص آپاندیسیت حاد در بیماران اطفال
کد: G-1756
نویسندگان: Alireza Ghattan ℗, Mohammad Matin Momeni, Arshia Bozorgnia , Hossein Ebrahimpour, Mohammad Shabani *
زمان بندی: زمان بندی نشده!
برچسب: سیستم های تصمیم یار بالینی
دانلود: دانلود پوستر
خلاصه مقاله:
خلاصه مقاله
Background and aims: Diagnosing acute appendicitis in children is challenging due to overlapping clinical presentations with other abdominal conditions. Delayed diagnosis can lead to complications, while misdiagnosis may cause unnecessary surgeries. This study develops and evaluates an AI-based decision tree model to improve diagnostic accuracy for acute appendicitis in children. Method: Data from 780 children suspected of appendicitis were analyzed, using clinical, laboratory, and imaging parameters from the Regensburg Pediatric Appendicitis Dataset on Kaggle. 702 cases were used for training, and 78 for evaluation. The model incorporated key diagnostic features, including Alvarado Score, Pediatric Appendicitis Score, WBC, CRP, RBC, appendix diameter, appendicolith presence, and clinical symptoms. Feature importance analysis identified the most influential variables in classification decisions. Results: The model achieved a training accuracy of 100.00% and a test accuracy of 92.31%. Precision was 0.92 for appendicitis and 0.93 for no appendicitis, with an overall weighted F1-score of 0.92. Key diagnostic contributors included appendix diameter, WBC count, neutrophil percentage, and appendicolith presence. These results indicate that the decision tree model can effectively differentiate between cases of appendicitis and non-appendicitis, reducing unnecessary surgeries and improving diagnostic precision Comparison with traditional scoring methods demonstrated the AI model’s superior accuracy. Conclusion: This study highlights the role of decision tree algorithms in improving the diagnosis of acute appendicitis in children. AI-driven decision-support systems may enhance efficiency and accuracy in clinical settings. However, dataset biases and the need for external validation remain challenges. Future research should explore ensemble learning, hybrid AI models, and real-world validation. Integrating this model into a clinical interface could further support physicians in emergency settings.
کلمات کلیدی
AcuteAppendicitis, ArtificialIntelligence, DecisionTree, MachineLearning, ClinicalDiagnosis, PediatricPatients
