NLP chatbots for pre-appointment dental symptom triage

Background and Aims: Possible delay in the care of dental patients is a common occurrence due to improper triage based on the severity of their symptoms or holding of patients at a similar acuity level. Recent evidence emphasizes the utility of Natural Language Processing (NLP) based chatbots for triaging symptoms. This review aims to evaluate compositional evidence as to the performance of transformer based NLP chatbots for pre-appointment dental symptom classification, primarily based on accuracy and workflow improving efficiencies. Results: The majority of studies demonstrated that BERT-based NLP chatbots could accurately classify dental symptoms, with typical performance metrics such as ۸۷–۸۹% precision and recall. Chatbots were particularly effective in identifying emergent cases such as abscesses and moderate pain. Post-deployment, clinics saw a ۱۰-۱۵% reduction in patient wait times and a significant increase in patient satisfaction, with ۶۸% of inquiries being handled autonomously, thus reducing staff workload. Methods: A systematic review, utilizing PubMed and Scopus (۲۰۱۵-۲۰۲۴), was conducted using potential keywords of 'NLP triage', 'chatbots in dentistry', 'AI in patient care'. Articles with references to assessing NLP models, particularly BERT based systems, used for classification of symptoms and triage in health care were included. Data were split to identify the types of symptoms such as tooth pain, bleeding, swelling, etc., different NLP models were trained to classify urgency into levels of emergent, urgent or routine levels of care. Important performance indicators included precision, recall, F۱-score and accuracy. Workflow efficiencies were assessed by comparing 'wait times' and satisfaction levels prior to using the chatbot and after its implementation in clinic. Conclusion: Transformer based NLP chatbots potentially provide an effective solution for dental symptom triage, enhancing patient centered care, easier clinic workflow efficiencies through time management and operational efficiencies. Using free-text with natural-language input for classification, is an innovative novel aspect of dental digital health.

کلیدواژه ها:

Patient-Centered Care ، Dental Workflow ، NLP Innovation