The focus of the article is on a generic approach to prediction of a healing process based on automatic measurement and modelling of mechanical disease indicators. As a specific example the healing process after treatment of advanced periodontitis is characterised and predicted based on measurement of tooth mobility (TM). For the measurement a new mechatronic instrument was developed that includes a pneumatically driven probe with incorporated sensors of tooth displacement and loading force. TM is measured by a computer controlled data acquisition system and predicted by a non-parametric regression. Similarly as in the characterisation of technical processes, a measured sample is described by TM time series detected on a particular patient, while the regression is determined by a set of samples obtained from a group of patients following periodontal therapy. The influence of surgical access therapy and systemic administration of metronidazole as an adjunct to the mechanical treatment is demonstrated. For this purpose the average changes of tooth mobility are determined on variously treated subgroups. A new method of healing process prediction is demonstrated and its performance is described by a prediction quality.