Machine Learning Model Predicts Postoperative Outcomes in Chronic Rhinosinusitis With Nasal Polyps.

Area of Science:

• Otolaryngology
• Medical Informatics
• Genomics

Background:

• Chronic rhinosinusitis with nasal polyps (CRSwNP) is a complex inflammatory condition.
• Predicting disease course after endoscopic sinus surgery (ESS) is challenging.
• Novel predictive tools are needed to optimize patient management.

Purpose of the Study:

• To evaluate the potential of Artificial Intelligence (AI) in predicting the disease trajectory of CRSwNP.
• To develop and validate machine learning algorithms for forecasting post-ESS outcomes.

Main Methods:

• Prospective study of 85 patients undergoing first ESS for nasal polyposis.
• Collection of preoperative data (demographics, blood eosinophils, endoscopy scores, SNOT-22, PHQ-9) and follow-up data.
• Development of Random Forest algorithms using clinical data and microRNAs (miR-125b, miR-203a-3p) to predict disease control, partial control, or relapse based on POSE scores.

Main Results:

• Machine learning models accurately predicted CRSwNP disease course at 18 months post-ESS.
• Prediction accuracy ranged from 69.23% (clinical data only) to 84.62% (including microRNAs).
• miR-125b significantly enhanced algorithm accuracy and was identified as a key predictive variable.

Conclusions:

• A machine learning algorithm shows promise for predicting the disease course in CRSwNP patients.
• Inclusion of microRNAs, particularly miR-125b, improves predictive accuracy.
• This AI approach may transform the prediction and management of CRSwNP.