A hybrid supervised and unsupervised machine learning approach for identifying nucleoside drugs using nanopore...

Area of Science:

• Biochemistry
• Analytical Chemistry
• Nanotechnology

Background:

• Nucleoside drugs are vital for treating cancer and viral infections.
• Current analysis methods like LC-MS/MS are complex, costly, and time-consuming.
• Next-generation detection methods are needed for efficient nucleoside drug analysis.

Purpose of the Study:

• To develop a portable, rapid, and precise high-throughput method for identifying nucleoside drugs.
• To enable simultaneous identification and differentiation of multiple nucleoside drugs.
• To overcome the limitations of existing analytical techniques for nucleoside drug detection.

Main Methods:

• Utilized a nanopore-based sensing platform for drug detection.
• Implemented a hybrid supervised and unsupervised machine learning workflow.
• Achieved single-molecule detection of nucleoside drugs with high accuracy.

Main Results:

• Demonstrated simultaneous identification of multiple anti-viral and anti-cancer nucleoside drugs.
• Achieved accurate drug identification from nanopore transmission readouts.
• Showcased the portability and efficiency of the developed method.

Conclusions:

• The proposed ML-assisted nanopore approach offers a next-generation solution for nucleoside drug analysis.
• This method enables rapid, precise, and high-throughput detection, overcoming previous analytical challenges.
• The technology holds promise for advancing clinical diagnostics and drug development.