Classification Tree Method for Determining Factors Associated with Halitosis

Authors : Mahmut KOPARAL, Utku Nezih YILMAZ, Ayşe ÖZCAN KÜÇÜK, Aydın KESKİNRÜZGAR, Fatih ÜÇKARDEŞ

Pages : 91-97

Doi:10.19127/bshealthscience.845961

View : 19 | Download : 9

Publication Date : 2021-05-01

Article Type : Research Paper

Abstract :Decision trees are data mining techniques for extracting hidden knowledge from large databases. This study was performed to establish the risk factors associated with halitosis by applying a decision tree model in a Turkish population and examining the interactions between these factors. We obtained data from a total of 1.290 patients, consisting of 645 patients with halitosis and 645 healthy controls. The subjects’ demographic characteristics, smoking status, alcohol intake, medical history and medications were assessed. The presence of potential intraoral causes of halitosis was determined by investigating perceived oral health problems such as caries, periodontal diseases, tongue coating, and oral cavity pathologies. Halitosis level was evaluated using an organoleptic scale. All data were subjected to classification tree analyses. Halitosis was significantly more common in patients with insert ignore into journalissuearticles values(80.9%); than without insert ignore into journalissuearticles values(20.7%); oral health problems insert ignore into journalissuearticles values(P < 0.001);. Halitosis was significantly less common in non-smokers without oral health problems than in smokers with oral health problems insert ignore into journalissuearticles values(14.5%; P < .001);. Halitosis was evident in all patients with oral health problems, smokers, and those with respiratory diseases insert ignore into journalissuearticles values(100%);. The effects of systemic diseases on halitosis were significant in non-smokers without oral health problems insert ignore into journalissuearticles values(P < 0.05);. Respiratory conditions showed significant effects on halitosis in smokers with oral health problems insert ignore into journalissuearticles values(P < 0.01);. We developed a decision tree model to identify risk factors associated with halitosis. The classification tree method showed that the most significant factors affecting halitosis were oral health problems followed by smoking status.
Keywords : Halitosis, Bad breath, Classification tree method, Data mining, Decision tree, Causes