Periodontitis is an irreversible disease leading to tooth loss, and 42% U.S. population suffers from periodontitis. Hence, diagnosing, monitoring, and determining its prevalence is critical to develop preventive strategies. However, a nationwide epidemiological study estimating the prevalence reported a concern about the discontinuation of such studies due to cost and ethical reasons. Therefore, this study determined the feasibility of utilizing electronic dental record (EDR) data and periodontitis case definition to automate periodontitis diagnosis. We utilized EDR data from the Indiana University School of Dentistry of 28,908 unique patients. We developed and tested a computer algorithm to diagnose periodontitis using the case definition. We found 44%, 22%, and 1% of patients with moderate, severe, and mild periodontitis, respectively. The algorithm worked with 100% sensitivity, specificity, and accuracy because of the excellent quality of the EDR data. We concluded the feasibility of providing automated periodontitis diagnosis from EDR data to conduct epidemiological studies across the US.
Read full abstract