Department of Chemistry, University of Bristol, Bristol, UK.
Magn Reson Chem. 2022 Nov;60(11):1087-1092. doi: 10.1002/mrc.5208. Epub 2021 Aug 30.
We demonstrate the potential for machine learning systems to predict three-dimensional (3D)-relevant NMR properties beyond traditional H- and C-based data, with comparable accuracy to density functional theory (DFT) (but orders of magnitude faster). Predictions of DFT-calculated N chemical shifts for 3D molecular structures can be achieved using a machine learning system-IMPRESSION (Intelligent Machine PREdiction of Shift and Scalar information Of Nuclei), with an accuracy of 6.12-ppm mean absolute error (∼1% of the δ N chemical shift range) and an error of less than 20 ppm for 95% of the chemical shifts. It provides less accurate raw predictions of experimental chemical shifts, due to the limited size and chemical space diversity of the training dataset used in its creation, coupled with the limitations of the underlying DFT methodology in reproducing experiment.
我们展示了机器学习系统在预测三维（3D）相关 NMR 属性方面的潜力，超越了传统的基于 H 和 C 的数据，并且具有与密度泛函理论（DFT）相当的准确性（但速度快几个数量级）。可以使用机器学习系统-IMPRESSION（智能机器预测核的位移和标量信息）来实现对 3D 分子结构的 DFT 计算的 N 化学位移的预测，其准确度为 6.12-ppm 平均绝对误差（约为 δ N 化学位移范围的 1%），并且对于 95%的化学位移，误差小于 20 ppm。由于用于创建它的训练数据集的大小和化学空间多样性有限，再加上基础 DFT 方法在再现实验方面的局限性，它对实验化学位移的原始预测不够准确。
