基于深度学习的老年人失牙风险预测及其移动端自助健康管理工具的构建

在当前我国老龄化现状下，如何充分激发老年人主动健康的巨大潜力，对老年人失牙及功能性失牙风险进行自助预测预警，并实现口腔健康自我管理，成为当下口腔健康领域的重要议题。国内外学者对于牙齿缺失相关风险预测做出相关研究，然而现有预测模型尚存在未反映我国老年人群特征、数据采集考虑不足、特异性和敏感性较低以及未形成后续风险评估或健康管理工具等诸多问题。为突破瓶颈、满足需求，本课题拟立足于我国老年人群特征，全面自助采集失牙（包括功能性失牙）相关风险因素，筛选核心风险因素并探究其流行病学作用机制；随后通过基于深度学习的多隐层神经网络算法精确拟合老年人失牙及功能性失牙风险预测模型，并检验模型的实际预警能力；最后将预测模型移植入移动端，添加个性化健康管理功能，构建简易的移动端自助口腔健康管理工具，真正实现以老年人个人或家庭为单位的失牙及功能性失牙风险自助预警和有针对性的自我健康管理，从而促进老年群体主动健康。

Due to the aging of the population in China, the active health promotion, self-prediction of tooth loss and functional tooth loss, and self-management among the elderly have become important issues in the field of oral health. Several researchers across the globe have investigated the risk assessment and prediction of tooth loss; however, there were quite a few limitations in the existing prediction models for tooth loss, such as not-based on the characteristics of older Chinese population, insufficient considerations for data collection, relatively low specificity and sensitivity, and the lack of development of relevant risk prediction or health management tools. In order to overcome the bottlenecks and meet demand, this project aims to focus on the characteristics of the elderly population in China, comprehensively collect the possible risk factors for tooth loss (including functional tooth loss) in a self-reported manner, determine the major risk factors, and explore the underlying epidemiological mechanisms; subsequently, we aim to derive accurate risk prediction models for tooth loss and functional tooth loss using the deep-learning-based multi-hidden-layer neural network and test their predictive ability; finally, we aim to integrate the prediction models with personalized health management measures to develop a simple mobile device-based self-management tool for oral health, which could achieve the self-warning and targeted self-management within individuals or families among the elderly and encourage them to take the initiative to safeguard and improve their own health.