Classification success of salivary interleukin-1β in periodontitis grading with artificial intelligencmodels: a cross-sectional observational study
Journal of Applied Oral Science, cilt.33, 2025 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 33
- Basım Tarihi: 2025
- Doi Numarası: 10.1590/1678-7757-2024-0580
- Dergi Adı: Journal of Applied Oral Science
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, MEDLINE, Directory of Open Access Journals
- Anahtar Kelimeler: Artificial intelligence, Biomarkers, Classification, Periodontit
- Süleyman Demirel Üniversitesi Adresli: Evet
Özet
Objectives: Finding certain biomarkers and threshold values of periodontitis and incorporating them into classifications can further highlight its impact on systemic health. This cross-sectional observational study aims to evaluate the efficacy of some biomarkers in grading periodontitis using artificial intelligence (AI) models. Methodology: AI models were developed to automatically classify periodontal status (N=240) and grades in periodontitis patients (n=120) using Python based on sociodemographic, anthropometric, clinical, radiological, and biochemical attributes. A total of 35 serum levels of whole blood attributes (white blood cell (WBC), platelet, erythrocyte, neutrophil, lymphocyte counts, and mean platelet volume), lipid profile [triglycerides; high-, low-, and very low-density lipoproteins (HDL, LDL, VLDL), and total cholesterol levels], salivary and serum interleukin (IL)-1β and matrix metalloproteinase (MMP)-8 levels), and 11 other attributes were used in the current classification. Results: In total, 23 out of 46 attributes achieved a 0.967 classification accuracy, whereas nine, a 0.858 classification accuracy. Attributes such as WBC, serum IL-1β, triglyceride/HDL ratio, neutrophil/lymphocyte ratio, and HDL were instrumental in periodontal status classification. HDL, LDL, neutrophil/lymphocyte ratio, total cholesterol, salivary IL-1β, and MMP-8 were key attributes in grading. Conclusions: AI models showed significant classification accuracy, particularly with serum and salivary IL-1β levels and other blood parameters, underscoring the potential of these biomarkers, which could be integrated into the current classification.