Analysis of Demographic and Comorbidity Data of Bangladeshi Female Dental Patients Towards Automatic Prediction of Comorbidities Risk

Timely prediction and accurate treatment of dental diseases considering other medical comorbidities is crucial. In greater context, analysis of demographic and comorbidity data associated with the dental diseases has great importance in the improvement of national healthcare through community awareness, computer assisted automatic prediction, and effective transdisciplinary medical treatment plan and management. In this study, we have collected and analyzed demographic and comorbidity data of Bangladeshi female dental patients with an aim to use the power and promises of data driven artificial intelligence towards digital transformation of dental healthcare. Total 1963 female patient's data with 63 attributes have been collected from three hospitals. However, thirteen dominant feature attributes were selected based on their statistical significance for machine learning based prediction. A rigorous analysis on the data has been conducted to investigate the dental health scenario. A Random Forest (RF) model is used to predict the dental disease from the comorbidity features with an emphasis of it's explainability in terms of the feature importance given by the model. The hyperparameters were tune using grid search method. From experimental results, HTN and Diabetes are found prominent comorbidity features and the RF prediction model exhibited an accuracy of 92.88% for 393 test data samples. The finding of this study might be useful for research and clinical applications in dental healthcare through awareness and computer aided prediction support.