Related Experiment Video
Updated: Aug 12, 2025
08:06
The Use of a β-lactamase-based Conductimetric Biosensor Assay to Detect Biomolecular Interactions
Published on: February 1, 2018
9.1K
Explainable Deep Learning-Assisted Photochromic Sensor for β-Lactam Antibiotic Identification
Xiaoqing Tan1, Yongtao Tang1, Tingting Yang1
1College of Chemistry and Materials Science, Guangdong Provincial Key Laboratory of Functional Supramolecular Coordination Materials and Applications, Su Bingtian Center for Speed Research and Training, Jinan University, Guangzhou 510632, China.
Analytical Chemistry
|January 30, 2023
Summary
This study uses explainable deep learning to analyze complex photochromic sensor data for identifying β-Lactams. The method achieves high accuracy, offering a transparent approach to understanding sensing mechanisms.
Area of Science:
- Analytical Chemistry
- Chemical Sensing
- Artificial Intelligence
Background:
- Photochromic sensors offer multi-analysis capabilities but present complexity.
- Deep learning (DL) excels at complex data but lacks transparency.
- Explainable DL addresses the 'black-box' nature of DL in sensing.
Purpose of the Study:
- To develop an explainable DL approach for photochromic sensing.
- To identify and quantify β-Lactams using a multi-state analysis array.
- To elucidate the photochromic sensing mechanism via transparent AI.
Main Methods:
- Spirooxazine metallic complexes used for a multi-state sensor array.
- Convolutional Neural Network (CNN) trained on 2520 fluorescence intensity images.
- Explainable AI techniques including molecular simulation and class activation mapping.
Main Results:
- Accurate discrimination of six β-Lactams with 97.98% prediction accuracy.
- Rapid quantification of β-Lactams in the range of 1–100 mg/L.
- Successful elucidation of the CNN model's decision-making process for sensor states.
Conclusions:
- Explainable DL provides a transparent and effective method for photochromic sensing.
- The approach enables accurate identification and quantification of analytes.
- This strategy aids in understanding complex sensing mechanisms for device optimization and discovery.
