Rapid identification of drug mechanisms is vital to the development and effective use of chemotherapeutics. Herein, we develop a multichannel surface-enhanced Raman scattering (SERS) sensor array and apply deep learning approaches to realize the rapid identification of the mechanisms of various chemotherapeutic drugs. By implementing a series of self-assembled monolayers (SAMs) with varied molecular characteristics to promote heterogeneous physicochemical interactions at the interfaces, the sensor can generate diversified SERS signatures for directly high-dimensionality fingerprinting drug-induced molecular changes in cells. We further train the convolutional neural network model on the multidimensional SAM-modulated SERS data set and achieve a discriminatory accuracy toward 99%. We expect that such a platform will contribute to expanding the toolbox for drug screening and characterization and facilitate the drug development process.
Keyphrases
- raman spectroscopy
- drug induced
- deep learning
- convolutional neural network
- liver injury
- gold nanoparticles
- sensitive detection
- loop mediated isothermal amplification
- adverse drug
- artificial intelligence
- induced apoptosis
- high throughput
- machine learning
- big data
- high resolution
- electronic health record
- oxidative stress
- genome wide
- cell cycle arrest
- quality improvement
- gene expression
- label free