A Deep Reinforcement Learning Approach to Droplet Routing for Erroneous Digital Microfluidic Biochips.

Area of Science:

• Biotechnology
• Microfluidics
• Artificial Intelligence

Background:

• Digital microfluidic biochips (DMFBs) offer compact and efficient biochemical experimentation for applications like DNA analysis and clinical diagnostics.
• The reliability of DMFBs is hindered by both detectable known errors and undetectable unknown errors that disrupt routing processes.
• Existing error management strategies struggle to address unknown errors effectively, limiting DMFB performance.

Purpose of the Study:

• To develop and evaluate a deep reinforcement learning-based routing algorithm for DMFBs.
• To enhance the reliability and success rate of routing processes in DMFBs by managing both known and unknown errors.
• To enable the detection of unknown errors during routing and identify optimal paths.

Main Methods:

• Implementation of a deep reinforcement learning (DRL) framework for DMFB routing.
• Development of an algorithm capable of identifying and adapting to both known and unknown error types during the routing process.
• Experimental validation comparing the DRL algorithm against existing routing methods.

Main Results:

• The DRL-based routing algorithm demonstrated superior performance compared to previous methods.
• The algorithm achieved higher success rates in routing scenarios involving both known and unknown errors.
• The proposed method successfully detected unknown errors during the routing process.
• The algorithm efficiently identified the most probable optimal routing paths.

Conclusions:

• Deep reinforcement learning offers a robust solution for enhancing DMFB reliability by addressing complex error types.
• The developed algorithm significantly improves routing success rates and error detection capabilities in DMFBs.
• This approach paves the way for more dependable and efficient microfluidic biochip applications.