Machine-learning-assisted characterization of peri-implant immune atlas improves peri-implantitis risk grading

Dental implants introduced exciting possibilities for functional reconstruction of the dentition. However, the emerging endemic of peri-implantitis affects over 25% of dental implants. A major challenge in managing peri-implantitis is the lack of a risk-grading system, which makes the maintenance recall schedules largely empirical. Clinical features alone fail to support informative stratification schemes. In addition, the pre-clinical models of peri-implantitis are built upon acute trauma-induced bone loss, which does not recapitulate the chronic course of peri-implantitis or take into account the interaction between host and keystone pathogens. Thus, we investigated a unique cohort of peri-implantitis to identify novel risk modifiers. To construct the landscape of immune infiltration at the peri-implant interface, we developed a robust outlier-resistant machine learning algorithm for immune deconvolution. This is an RNA-Seq data set built upon peri-implant granulation tissue from 24 patients with per-implantitis.