QSPR models for sublimation enthalpy of energetic compounds

Prediction of Drug-like Compounds Solubility in Supercritical Carbon Dioxide: A Comparative Study between Classical Density Functional Theory and Machine Learning Approaches

“…Turning to the computation of the solubility with the help of eq one also needs to determine the saturation vapor pressure of the solute under investigation. Due to the fundamental difficulties in fast and accurate theoretical calculation of the sublimation thermodynamics, − we turn to the literature experimental data on temperature dependence of the saturation vapor pressure, when it is available. In the opposite case, one must resort to approximation techniques.…”

Bionic inspired multifunctional modular energetic materials: an exploration of new generation of application-oriented energetic materials

“…With the development of computational level, theoretical chemistry and high-throughput screen inject new vitality into the search for energetic materials under the consideration of cost and safety. 13–15 During the high-throughput screen, machine learning offers a vital tool for effective performance prediction, such as density, 16–18 heat of formation, 19,20 detonation properties, 21–23 and decomposition temperature. 24–27 We have explored to some extent in this field as well: in 2021, we conducted a domain-related knowledge-promoted high-throughput cage scaffold screening from the ZINC15 database containing over 130 000 scaffolds and merged it with a combinatorial design to alleviate the lack of cage energetic materials.…”

MDs-NP: a property prediction model construction procedure for naphtha based on molecular dynamics simulation

“…QSPR is a method for building machine learning models to mine the relationships between the properties and structure of a molecule. As a new field in the natural science, researches on QSPR include predicting the physicochemical behaviors and properties (boiling point [13], toxicity [14], heat capacity [15], sublimation enthalpy [16]) of a molecule based on structural information, identifying potential candidates with specified properties or functionalities [17]. Generally, a QSPR model uses numerical molecular structural characteristics as input, achieving efficient as well as accurate calculations and predictions of molecular properties based on various machine learning algorithms [17][18][19].…”

“…Data from experimental measurements are generally recorded in databases or handbooks, and are the most frequently used in establishing QSPR models. Whilst data from experiments have the highest accuracy, they are time-consuming and costly [16], and only include some of the experimental properties of single molecules at a few temperature and pressure points, missing data of complex organic systems. Data from theoretical calculations are largely guided by experience and intuition, using empirical formulas for properties, and are often used for rapid property calculations for single molecules [21,23].…”