We present the results of creating models based on machine learning methods aimed at accurately predicting absorption maximum wavelength and half maximal inhibitory concentration of BODIPY-appended platinum complexes (Pt-BODIPYs). For this purpose, we compiled a unique database on spectral characteristics and cytotoxicity parameters of Pt-BODIPYs, comprising 447 records. Using this database and applying the CatBoost/ECFP and kNN/text vectorization algorithms, while considering various experimental conditions (solvent, cell line, presence or absence of light), we successfully trained models to forecast the absorption maximum wavelength (MAE ∼ 14 nm) and half maximal inhibitory concentration (MAE ∼ 0.26 log unit) for Pt-BODIPYs. For enhanced accessibility, we have made these trained models available on our web service ChemPredictor at http://chem-predictor.isc-ras.ru/individual/abs_log/.
