Deep Learning and Selective Classification Approaches for Predicting Hospitalization Duration and Severity Classification in Acute Pancreatitis Patients Akut Pankreatit Hastalarinda Yati s S reci Tahminive Siddet Siniflandirmasi zerine Derin grenme veSe imli Siniflandirma Yakla simlari

Deveci M., Akkilic Z. N., Dokur S., Dokur M., Sezer A.

2025 Innovations in Intelligent Systems and Applications Conference, ASYU 2025, Bursa, Türkiye, 10 - 12 Eylül 2025, (Tam Metin Bildiri) identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/asyu67174.2025.11208262
  • Basıldığı Şehir: Bursa
  • Basıldığı Ülke: Türkiye
  • Anahtar Kelimeler: Acute Pancreatitis, Balthazar Score, Classification, Feature Selection, Length of Stay, Machine Learning, Regression Analysis
  • İstanbul Yeni Yüzyıl Üniversitesi Adresli: Evet

Özet

Acute pancreatitis is a disease that can lead to serious complications and requires early diagnosis and appropriate treatment planning for optimal management. This study aims to predict the length of hospital stay and to classify the severity of the disease in patients with acute pancreatitis. A large dataset including patients' clinical data and laboratory analyses was analyzed within the scope of the study. For predicting the length of stay, the Random Forest regression model was utilized and achieved an R2 score of 0.61. For disease severity classification, Support Vector Machine, Random Forest, and Multilayer Perceptron (MLP) algorithms were compared, reaching up to %70 classification success. Feature selection was performed using Chi-Squared test, Mutual Information, and Lasso regularization methods to identify the most significant parameters. The results demonstrate that machine learning approaches can provide valuable decision support to clinicians in the management of pancreatitis cases.