The model predicts the solubility of drugs in supercritical carbon dioxide by employing the CatBoost method with CDK descriptors and the drug’s melting point as input variables. Users need to input the drug’s representation in the SMILES format, in addition to the supercritical state’s temperature and pressure conditions.

These models are capable of predicting the absorption maximum (CatBoost/ECFP) and IC₅₀ (kNN/text vectorization) values  for BODIPY-appended platinum complexes. Additionally, the models facilitate the search for similar structures by calculating the Euclidean distance between the Morgan Fingerprint vectors of the input structure and the structures in the training database.

This model is based on the CatBoost method using the Morgan fragment descriptors. The complete experimental dataset consisted of 10,849 records, which were used to evaluate the absorption maximum wavelength of porphyrins and metalloporphyrins. The dataset included both free porphyrins and chlorins, as well as their complexes with various metal ions such as zinc(II), nickel(II), copper(II), cobalt(II), gold(III), indium(III), palladium(II), platinum(II), silver(II), cadmium(II), magnesium(II), and aluminum(III).

This model can predict the absorption maximum and molar absorption coefficient (CatBoost/ECFP) for various classes of dyes, such as xanthenes, acridines, diarylmethanes, anthraquinones, and dipyrromethenes.

This model based on an RFR machine learning method using the ISIDA fragment descriptors for predicting the ¹¹B NMR chemical shift of BODIPY.