通过SHAP和特征分析解释改进机器学习分类预测
Leonardo Bernal1,2, Giulio Rastelli1, Luca Pinzi1
1Department of Life Sciences, University of Modena and Reggio Emilia, Via Giuseppe Campi 103, 41125 Modena, Italy.
Journal of chemical information and modeling
|October 20, 2025
概括
这项研究引入了一种新的方法,将SHapley添加式解释 (SHAP) 与特征分析相结合,以提高机器学习模型在药物发现中的准确性. 该方法有效地识别和标记错误分类的化合物,提高虚拟查的预测性能.
科学领域:
- 计算化学和化学信息学
- 机器学习在药物发现中的作用
- 癌症研究 癌症研究
背景情况:
- 基于树的机器学习 (ML) 算法,如额外树 (ET),随机森林 (RF),梯度增强机 (GBM) 和XGBoost (XGB) 在早期药物发现中至关重要.
- 这些模型经常面临错误分类和有限的解释性挑战,阻碍了实际应用.
- 沙普利添加式解释 (SHAP) 提供了一种方法来理解特征的重要性,并可能改善模型预测.
研究的目的:
- 开发和验证一种新的方法,整合SHAP值和特征分析,以减少ML模型中的错误分类错误.
- 使用前列腺癌细胞系数据对ET,RF,GBM和XGB算法的性能进行基准测试.
- 创建一个错误分类检测框架,以提高虚拟查预测的可靠性.
主要方法:
- 使用RDKit和ECFP4分子描述器对ET,RF,GBM和XGB分类器进行基准测试.
- 应用SHAP价值分析来理解预测驱动因素并识别错误分类的化合物.
- 开发和测试四种错误分类检测过规则:RAW,SHAP,RAW OR SHAP,以及RAW AND SHAP.
主要成果:
- 在PC3,LNCaP和DU-145细胞系的抗增殖活性数据上,GBM和XGB模型实现了高性能 (MCC>0.58,F1得分>0.8).
- SHAP分析显示,错误分类的化合物往往具有相反类的典型特征值.
- "RAW OR SHAP"规则成功地发现了大量错误分类的化合物 (在LNCaP中高达63%).
结论:
- 拟议的SHAP和特征分析的整合提供了一个有效的策略,以检测和减轻ML模型中的错误分类.
- 开发的过规则通过排除可能错误的预测来提高分类器的性能.
- 这种方法为提高药物发现中虚拟查的准确性和可靠性提供了有价值的工具.
相关概念视频
01:19Role of Shaping in Operant Conditioning
956
Shaping is a technique used in operant conditioning to train complex behaviors by rewarding successive approximations toward the target behavior. This method is necessary because organisms are unlikely to perform complex behaviors spontaneously. Instead, shaping breaks down the desired behavior into small, manageable steps.
The steps involved in shaping begin with reinforcing any response that resembles the desired behavior. For example, parents might praise a child for picking up one toy. As...
The steps involved in shaping begin with reinforcing any response that resembles the desired behavior. For example, parents might praise a child for picking up one toy. As...
956
01:30Classification of Signals
1.3K
In signal processing, signals are classified based on various characteristics: continuous-time versus discrete-time, periodic versus aperiodic, analog versus digital, and causal versus noncausal. Each category highlights distinct properties crucial for understanding and manipulating signals.
A continuous-time signal holds a value at every instant in time, representing information seamlessly. In contrast, a discrete-time signal holds values only at specific moments, often denoted as x(n), where...
A continuous-time signal holds a value at every instant in time, representing information seamlessly. In contrast, a discrete-time signal holds values only at specific moments, often denoted as x(n), where...
1.3K
01:11Regression Analysis
8.0K
Regression analysis is a statistical tool that describes a mathematical relationship between a dependent variable and one or more independent variables.
In regression analysis, a regression equation is determined based on the line of best fit– a line that best fits the data points plotted in a graph. This line is also called the regression line. The algebraic equation for the regression line is called the regression equation. It is represented as:
In regression analysis, a regression equation is determined based on the line of best fit– a line that best fits the data points plotted in a graph. This line is also called the regression line. The algebraic equation for the regression line is called the regression equation. It is represented as:
8.0K
01:11Statistical Analysis: Overview
14.3K
When we take repeated measurements on the same or replicated samples, we will observe inconsistencies in the magnitude. These inconsistencies are called errors. To categorize and characterize these results and their errors, the researcher can use statistical analysis to determine the quality of the measurements and/or suitability of the methods.
One of the most commonly used statistical quantifiers is the mean, which is the ratio between the sum of the numerical values of all results and the...
One of the most commonly used statistical quantifiers is the mean, which is the ratio between the sum of the numerical values of all results and the...
14.3K