3D-QSAR (three-dimensional quantitative structure-activity relationship) technology plays a key role in computational chemistry, its main goal is to predict the activity and other structure-related properties of molecules in organisms. This technique is based on the central premise that there is a quantifiable correlation between the mathematical characterization of molecular structure and its biological activity. In the application of 3D-QSAR, chemical descriptors (such as molecular configuration and physicochemical properties) are used as the basis for analysis, which are then combined with the bioactive parameters of the molecule (such as efficacy and toxicity). The 3D-QSAR method can predict the biological activity of unknown compounds by employing a variety of mathematical models, such as linear regression, support vector machines or neural networks. This technology has demonstrated its value in many fields such as drug development, environmental risk assessment and agricultural chemical development, providing scientists with an efficient tool that not only speeds up the screening and design process for specific bioactive compounds, but also reduces experimental costs and dependence on animal experiments.

In the 3D-QSAR modeling process, we investigated the three-dimensional structural characteristics of 20 drug molecules, including ALogP, molecular weight, number of rings (including aromatic rings), number of hydrogen bond donors and acceptors, number of flexible bonds, polarized surface area of molecular fragments, torsional energy, solvent accessible surface area, molecular volume, and other relevant molecular fingerprint information. These characteristics relate to the number of times a drug molecule is solubilized by a particular solvent. Our model successfully predicts the solubilization factor of a drug molecule in a specific solubilizer, and the error between the predicted results and experimental data and other simulated values is within 5%, demonstrating the accuracy and reliability of the model.