Longevity Recommender Model for Root Canal Treatment using Fusion Deep Learning Algorithm

Abstract

Despite having the high success rate of endodontic therapy, many patients still experience complications in root canal treatment. The reason that has been observed are a wide range of clinical and non-clinical factors. So, it is very important to avoid or at least cut down on the most common reasons which are responsible for root canal treatment failure. Therefore, identifying these causes helps to take the necessary treatments. Here, machine learning and deep learning techniques are employed to identify the non-clinical and clinical causes of the RCT failure. Consequently, using logistic regression on textual data offers significant accuracy in detection of non-clinical along with some clinical causes of RCT failure such as age, oral hygiene, tooth location etc. Again, using a convolutional neural network (CNN) also effectively detects the clinical cause such as overfilling, under filling, perforation, or root resorption. Moreover, the fusion of logistic regression and CNN help to predict the longevity of the treatment with the accuracy 91.27%, precision 93.55%, sensitivity 93.12% , specificity 87.72% for class 0(low class) . also, it predicts class 1(high class) with accuracy 91.27%, precision 86.96%, sensitivity 87.72% and specificity 93.12%. newlineKeywords: Root Canal Treatment Failure, Machine learning approached, Deep learning approaches, Toot Longevity Prediction, CNN, Overfilling, Under Filling, Perforation, Logistic Regression, Fusion Approach newlineIII newline