Differentiation of Clinical Patterns Associated With Rhinologic Disease.

Area of Science:

• Rhinology
• Artificial Intelligence
• Data Science

Background:

• Common rhinologic diagnoses often present with overlapping symptoms.
• Identifying distinct patterns in clinical data can aid early diagnosis and outcome prediction.

Purpose of the Study:

• To assess the feasibility of using artificial intelligence (AI) to analyze patient data.
• To develop clinically meaningful diagnostic groups for rhinologic conditions based on data patterns.

Main Methods:

• A cross-sectional study analyzed prospectively acquired data from 545 patients at a tertiary rhinology clinic.
• Data included nasal endoscopy findings, patient-reported quality of life (PRQOL) scores, peripheral eosinophil counts, and medical history.
• Unsupervised non-hierarchical cluster analysis was applied to 22 input variables.

Main Results:

• Seven unique and clinically relevant patient clusters were identified, primarily driven by PRQOL scores and demographics.
• Chronic rhinosinusitis without nasal polyposis (CRSsNP) clusters showed low asthma frequency and eosinophil counts.
• Chronic rhinosinusitis with nasal polyposis (CRSwNP) clusters were linked to high asthma rates, elevated eosinophils, and specific symptom scores (NOSE, SNOT-22).
• Allergic rhinitis (AR) appeared across multiple clusters, while Allergic Rhinosinusitis (RARS) was associated with younger patients.

Conclusions:

• AI-driven analysis of readily available clinical data, including PRQOL and demographics, can enhance diagnostic efficiency for rhinologic conditions.
• This approach may reduce the need for ancillary studies in diagnosing and managing sinonasal diseases.