Title : AI-powered detection of functional gastrointestinal disorders in adults visiting tertiary hospitals
Abstract:
Background: Functional gastrointestinal disorders (FGIDs), such as irritable bowel syndrome and functional dyspepsia, are commonly encountered in tertiary care settings and are often challenging to diagnose due to the absence of structural abnormalities. Artificial intelligence (AI) offers a promising avenue to enhance diagnostic accuracy by analyzing complex symptom patterns and clinical data.
Objective: To examine the diagnostic accuracy of AI tools in identifying functional gastrointestinal disorders among adult patients in tertiary hospitals in wah cantt and Rawalpindi, Pakistan.
Methods: A cross-sectional study was conducted over eight months, enrolling 420 adult patients presenting with gastrointestinal symptoms at three tertiary hospitals. Participants underwent evaluation based on Rome IV criteria, which served as the clinical diagnostic reference. A supervised AI model incorporating machine learning and natural language processing was applied to analyze structured and unstructured clinical data. Diagnostic performance was assessed using sensitivity, specificity, predictive values, and area under the curve (AUC). Cohen’s kappa coefficient was used to evaluate agreement between AI and clinical diagnoses. Data analysis was performed using SPSS v26.0.
Results: The AI tool demonstrated a sensitivity of 88.2%, specificity of 84.5%, and overall diagnostic accuracy of 87.1%. The AUC for overall FGID classification was 0.912, with highest performance in diagnosing functional dyspepsia (AUC 0.914) and irritable bowel syndrome (AUC 0.902). Strong diagnostic agreement was observed between AI and physician diagnoses (κ = 0.83). The findings support AI's reliability in replicating expert-level FGID classification in clinical settings.
Conclusion: AI tools can effectively support the diagnosis of FGIDs in tertiary care, offering a valuable adjunct to clinical judgment and enhancing consistency in complex patient presentations.
Keywords: Artificial Intelligence, Diagnosis, Functional Colonic Diseases, Functional Gastrointestinal Disorders, Gastroenterology, Machine Learning, Natural Language Processing
