Periodontal disease is a serious gum infection that can lead to tooth loss. Understanding the risk factors for tooth loss and building clinical prediction models for future tooth loss have been continuing efforts among oral health researchers. However, previous studies have only used covariates obtained from baseline to estimate long-term tooth loss despite the availability of routinely collected data. Therefore, we illustrated the use of joint models for longitudinal and survival data to estimate risk factors for tooth loss as a function of time-varying endogenous periodontal biomarkers. Through a simulation study and application to a longitudinal study of dental disease, we showed that the joint model can incorporate time-varying periodontal biomarkers to accurately estimate the hazard of tooth loss. In this talk, I will first present the work with technical details aimed towards a statistical audience and later present the same work aimed towards a non-statistical audience.