Prediction of Chronic Periodontitis Severity Using Machine Learning Models Based On Salivary Bacterial Copy Number

Abstract

Periodontitis is a widespread chronic inflammatory disease caused by interactions between periodontal bacteria and homeostasis in the host. We aimed to investigate the performance and reliability of machine learning models in predicting the severity of chronic periodontitis. Mouthwash samples from 692 subjects (144 healthy controls and 548 generalized chronic periodontitis patients) were collected, the genomic DNA was isolated, and the copy numbers of nine pathogens were measured using multiplex qPCR. The nine pathogens are as follows: Porphyromonas gingivalis (Pg), Tannerella forsythia (Tf), Treponema denticola (Td), Prevotella intermedia (Pi), Fusobacterium nucleatum (Fn), Campylobacter rectus (Cr), Aggregatibacter actinomycetemcomitans (Aa), Peptostreptococcus anaerobius (Pa), and Eikenella corrodens (Ec). By adding the species one by one in order of high accuracy to find the optimal combination of input features, we developed an algorithm that predicts the severity of periodontitis using four machine learning techniques. The accuracy was the highest when the models classified “healthy” and “moderate or severe” periodontitis (H vs. M-S, average accuracy of four models: 0.93, AUC = 0.96, sensitivity of 0.96, specificity of 0.81, and diagnostic odds ratio = 112.75). One or two red complex pathogens were used in three models to distinguish slight chronic periodontitis patients from healthy controls (average accuracy of 0.78, AUC = 0.82, sensitivity of 0.71, and specificity of 0.84, diagnostic odds ratio = 12.85). Although the overall accuracy was slightly reduced, the models showed reliability in predicting the severity of chronic periodontitis from 45 newly obtained samples. Our results suggest that a well-designed combination of salivary bacteria can be used as a biomarker for classifying between a periodontally healthy group and a chronic periodontitis group.

Fusobacterium nucleatum Acts as a Pro-carcinogenic Bacterium in Colorectal Cancer: From Association to Causality

“…As a symbiotic bacterium, F. nucleatum serves as a structural support for other bacteria to form the oral biofilms, which are essential for the normal oral microenvironment ( Lamont et al, 2018 ; Zhang et al, 2018 ). On the other hand, since it has been isolated from clinical infections and multiple tumor samples, such as periodontitis ( Kim E. H. et al, 2020 ), adverse pregnancy ( Vander Haar et al, 2018 ; Figuero et al, 2020 ), appendicitis ( Hattori et al, 2019 ), CRC ( Castellarin et al, 2012 ; Kostic et al, 2012 ), and breast cancer ( Parhi et al, 2020 ), it has been regarded as an opportunistic pathogen and a tumor-associated bacterium. To further explore its mechanisms to promote CRC, we first introduce four basic biological characteristics associated with pathogenicity.…”

Microbiome of Saliva and Plaque in Children According to Age and Dental Caries Experience

“…Each feature was evaluated based on its importance in each group using random forest models (Table S1). The features were added one by one in order of importance from the highest to the lowest, resulting in many models with various feature combinations (Table S2), as previously done by Kim et al [41]. The models with the best accuracy were selected.…”

Deep learning in periodontology and oral implantology: A scoping review

“…The fourth application was the prediction of treatment outcomes, for example, for implants 56‐58 and using predictors like trabeculae microstructure parameters, 57 insertion torque curve, 56 patients' health condition, 58 or for periodontally affected teeth 59 . For the latter, a range of clinical parameters has been employed, 60,61 while biosample data like saliva has only been infrequently used 62 . Again, models which combine the wealth of social and routinely collected data (on probing pocket depths, attachment level, mobility, etc.…”

“…), as well as further biomarkers could be beneficial, using the complex associations between different predictors 63,64 . Moreover, the dynamic nature of periodontal disease should be reflected, and longitudinally available data may be employed in time series for temporal training models 21,62 . The prediction of tooth loss as a major outcome of (by large untreated) periodontitis has been attempted by a range of (not deep) learning models—with limited success so far 65,66 …”

“…59 For the latter, a range of clinical parameters has been employed, 60,61 while biosample data like saliva has only been infrequently used. 62 Again, models which combine the wealth of social and routinely collected data (on probing pocket depths, attachment level, mobility, etc. ), as well as further biomarkers could be beneficial, using the complex associations between different predictors.…”