Can Artificial Intelligence Revolutionise Surgical Decision-Making for Appendectomy? A Narrative Review

Area of Science:

• Medical Imaging
• Computational Biology
• Surgical Diagnostics

Background:

• Acute appendicitis is a frequent cause of acute abdomen, yet misdiagnosis and negative appendectomies persist.
• Advancements in Artificial Intelligence (AI), including machine learning (ML) and deep learning, offer potential solutions for diagnostic challenges.

Purpose of the Study:

• To review the literature on AI's application in diagnosing and prognosing acute appendicitis.
• To assess the efficacy of various AI models compared to traditional diagnostic scores.

Main Methods:

• Conducted a literature review of studies on AI for appendicitis diagnosis and prognosis.
• Included studies from PubMed and Cochrane databases, excluding those solely based on clinical or radiology reports.

Main Results:

• AI models, particularly random forest (RF), demonstrated high diagnostic accuracy, surpassing traditional scores like the Alvarado score.
• Machine learning methods (SVM, XGBoost) effectively predicted prognosis and identified complicated appendicitis cases.
• Neural networks (NN) and logistic regression (LR) also showed promising results in diagnosis.

Conclusions:

• AI holds significant potential to improve acute appendicitis diagnosis and prognosis.
• Addressing data requirements and standardization is crucial for widespread clinical adoption of AI in appendicitis management.