Oral health problems are closely associated with the analysis of dental tissue changes and the stomatologic treatment that follows. This paper explores the use of diffuse reflectance spectroscopy in the detection of dental tissue disorders. The data set includes 343 measurements of teeth spectra in the wavelength range from 400 to 1700 nm. The proposed methodology focuses on computational and statistical methods and the use of these methods for the classification of dental tissue into two classes (healthy and unhealthy) by estimating the probability of class membership. Signal processing is based on the difference between the healthy and unhealthy teeth reflectance spectra in the infrared and visible ranges. Selected features associated with observed spectra are then used for machine learning classification based on the experience of an expert in stomatology during the learning stage. The proposed modification of the weighted k-nearest neighbour method provides class boundaries and the probability of class membership during the verification stage. The accuracy of the classification process reached 95.4%. The proposed methodology and graphical user interface point to the possibility of using absorption spectroscopy in the evaluation of tissue quality changes and its possible implementation in the clinical environment.