Development of Machine Learning Models for Ion-Selective Electrode Cation Sensor Design

Abstract

Polyvinyl chloride (PVC) membrane-based ion-selective electrode (ISE) sensors are common tools for water assessments, but their development relies on time-consuming and costly experimental investigations. To address this challenge, this study combines machine learning (ML), Morgan fingerprint, and Bayesian optimization technologies with experimental results to develop high-performance PVC-based ISE cation sensors. By using 1745 data sets collected from 20 years of literature, appropriate ML models are trained to enable accurate prediction and a deep understanding of the relationship between ISE components and sensor performance (R ² = 0.75). Rapid ionophore screening is achieved using the Morgan fingerprint based on atomic groups derived from ML model interpretation. Bayesian optimization is then applied to identify optimal combinations of ISE materials with the potential to deliver desirable ISE sensor performance. Na⁺, Mg²⁺, and Al³⁺ sensors fabricated from Bayesian optimization results exhibit excellent Nernst slopes with less than 8.2% deviation from the ideal value and superb detection limits at 10^-7 M level based on experimental validation results. This approach can potentially transform sensor development into a more time-efficient, cost-effective, and rational design process, guided by ML-based techniques.

Figure 1

Diagram of the ML-Bayesian optimization framework for ion-selective electrode cation sensor design.

Figure 2

Shapley additive explanation (SHAP) plot interprets the models’ fabrication condition contributions to the Nernst slope in the data set. The X-axis is the SHAP value, and the positive values indicate that the Nernst slope can be increased by the specific features listed on the left Y-axis. Meanwhile, the negative values indicate a reduction in the Nernst slope. The size of each feature’s value is colored from blue to red, corresponding to the smallest and largest values (in the charge states, blue is +1, purple is +2, and red is +3). The pattern for each feature is composed of small dots, and each dot represents a sample containing this feature.

Figure 3

Atomic groups serve as (a) positive contributors and (b) negative contributors to the Nernst slope. The unlabeled blue circles represent carbon atoms. Each feature number denotes the feature position in the Morgan fingerprint vector. The blue circles are atoms. The gray lines represent the bonds that are not included in the features. The asterisks (*) represent other atoms that are connected.

Figure 4

Identification of optimal combinations from Bayesian optimization.