Predicting the success of dental implant treatment using Artificial intelligence and Machine learning

Background and aims: Nowadays, dental implants are widely used to replace missing teeth.However, complications, failures, and diseases are not uncommon. Since the price of this treatmentis higher than conventional methods, its success rate should be carefully evaluated beforetreatment. Accurate prediction of implant success is influenced by many factors, making it achallenging subject for dentists. One of the significant services that machine learning (ML) hasprovided to the medical context is the prediction of treatment outcomes. We aimed to review studiesthat used prediction models based on artificial intelligence (AI) and ML for predicting dentalimplant treatment success.Method and Materials: Our review question focused on the literature search about methods ofAI-assisted implant treatment success prediction. Keywords consisted of artificial intelligence,machine learning, dental implants, prognosis, and prediction. An electronic search was conductedin ۵ databases: MEDLINE/PubMed, EMBASE, Web of Science, Cochrane, and Scopus. Amanual search was conducted too. Studies that investigated the clinical applications of artificialintelligence to predict implant success using patient risk factors were included. Review articles,letters to the editors, and posters were excluded. Relevant articles published up to October ۲۰۲۲were identified, quality assessed, and data extracted by two reviewers.Result: At first, a duplication check was conducted, and ۷۴ articles were identified. After screeningand applying inclusion and exclusion criteria, seven studies were reviewed that investigatedthe success of implant treatment with AI methods. Different ML models were used in these studies:support vector machine (SVM), artificial neural network (ANN), logistic regression (LR),random forest (RF), decision tree (DT), ensemble selection (ES), K-Nearest Neighbors (K-NN),and Naïve Bayes. Various studies used different input data for predicting implant treatment success,such as demographics, physical condition, lifestyle, surgeon background, anatomic condition,surgical information, implant attributes, and prosthetics attributed. Among the includedstudies, the accuracy, sensitivity, and specificity of these methods in all included studies rangedfrom ۶۲.۴۰% (for LR) to ۹۹.۲۵% (for NN), ۴۸.۰۸% (for K-NN) to ۹۷.۶۳% (for NN), and ۶۱.۱۱%(for DT) to ۱۰۰% (for SVM), respectively.Conclusion: According to this review, NN method showed the highest accuracy and sensitivity,and the highest specificity belonged to SVM. They also reported that dental implant prognosismostly depends on factors such as the mesio-distal position of the inserted implant, fixture width,and implant system. Based on this review, AI models have great potential for implant successprediction. However, the use of AI and ML has not been integrated into routine dentistry. AI isstill in the research phase, and further studies are required to assess the clinical performance ofthese methods in dentistry.

کلیدواژه ها:

artificial intelligence ، dentistry ، machine learning ، dental implants ، Implant prognosis