A Machine Learning Approach for Predicting Defluorination of Per- and Polyfluoroalkyl Substances (PFAS) for Their Efficient Treatment and Removal
A Machine Learning Approach for Predicting Defluorination of Per- and Polyfluoroalkyl Substances (PFAS) for Their Efficient Treatment and Removal
复制标题
DOI：
10.1021/acs.estlett.9b00476
复制
发表时间：
2019-10-01
影响因子：
10.9
通讯作者：
Wong, Bryan M.
中科院分区：
文献类型：
--
作者：
Raza, Akber;Bardhan, Sharmistha;Wong, Bryan M.
我们提出了机器学习在全氟和多氟烷基物质（PFAS）上的第一个应用，用于预测和合理化碳氟（C-F）键解离能，以帮助它们的有效处理和去除。使用各种机器学习算法（包括随机森林，最小绝对收缩和选择算子回归，以及前馈神经网络），我们能够获得非常准确的C-F键离解能预测（偏差小于0.70 kcal/mol），在PFAS参考数据的化学精度范围内。此外，我们证明了我们的机器学习方法非常高效，只需不到10分钟的时间来训练数据，不到一秒钟就可以预测新化合物的C-F键解离能。最重要的是，我们的方法只需要了解PFAS结构中简单的化学连通性就可以产生可靠的结果，而无需求助于计算成本高昂的量子力学计算或三维结构。最后，我们提出了一种无监督机器学习算法，该算法可以自动分类和合理化PFAS结构中的化学趋势，否则很难人工可视化或手动处理。总的来说，这些研究(1)包括机器学习技术在PFAS结构中的首次应用，以预测/理顺C-F键离解能；(2)在协助实验人员对日益复杂的PFAS结构（或其他未知环境污染物）中的特定键进行定向除氟方面显示出巨大的希望。
We present the first application of machine learning on per- and polyfluoroalkyl substances (PFAS) for predicting and rationalizing carbon-fluorine (C-F) bond dissociation energies to aid in their efficient treatment and removal. Using a variety of machine learning algorithms (including Random Forest, Least Absolute Shrinkage and Selection Operator Regression, and Feed-forward Neural Networks), we were able to obtain extremely accurate predictions for C-F bond dissociation energies (with deviations less than 0.70 kcal/mol) that are within chemical accuracy of the PFAS reference data. In addition, we show that our machine learning approach is extremely efficient, requiring less than 10 min to train the data and less than a second to predict the C-F bond dissociation energy of a new compound. Most importantly, our approach only needs knowledge of the simple chemical connectivity in a PFAS structure to yield reliable results-without recourse to a computationally expensive quantum mechanical calculation or a three-dimensional structure. Finally, we present an unsupervised machine learning algorithm that can automatically classify and rationalize chemical trends in PFAS structures that would otherwise have been difficult to humanly visualize or process manually. Collectively, these studies (1) comprise the first application of machine learning techniques for PFAS structures to predict/rationalize C-F bond dissociation energies and (2) show immense promise for assisting experimentalists in the targeted defluorination of specific bonds in PFAS structures (or other unknown environmental contaminants) of increasing complexity.