Deep neural network provides robust detection of disease biomarkers in real time

Holger Schmidt, distinguished professor of electrical engineering and computer science at UC Santa Cruz, and his group have long focused on developing unique and highly sensitive devices called optofluidic chips to detect biomarkers.

Schmidt’s graduate student Vahid Ganjalizadeh led an effort to use machine learning to improve his systems by improving their ability to accurately classify biomarkers. The deep neural network he developed classifies particle signals with 99.8% accuracy in real time, in a system that is relatively cheap and portable for point-of-care applications, as shown in a new paper in Scientific reports from nature.

When taking biomarker detectors into the field or to a point of care such as a health clinic, the signals received by the sensors may not be of as high quality as those in a laboratory or controlled environment. This can be due to a variety of factors, such as the need to use cheaper chips to reduce costs, or environmental characteristics such as temperature and humidity.

To address the challenges of a weak signal, Schmidt and his team developed a deep neural network that can identify the source of that weak signal with great confidence. The researchers trained the neural network with known training signals, teaching it to recognize potential variations it might see, so it can recognize patterns and identify new signals with very high precision.

First, a parallel cluster wavelet analysis (PCWA) approach designed in Schmidt’s lab detects that a signal is present. The neural network then processes the potentially weak or noisy signal, identifying its source. This system works in real time, so users can receive results in a fraction of a second.

“It’s all about making the most of low-quality signals, and doing it really quickly and efficiently,” Schmidt said.

A smaller version of the neural network model can be run on portable devices. In the article, the researchers run the system on a Google Coral Dev board, a relatively inexpensive edge device for accelerated execution of AI algorithms. This means that the system also requires less power to run the processing compared to other techniques.

“Unlike some research that requires supercomputers to run to perform high-precision detection, we show that even a compact, portable and relatively inexpensive device can do the job for us,” Ganjalizadeh said. “It makes it available, feasible and portable for point-of-care applications.”

The entire system is designed to be used entirely locally, which means that data processing can be done without internet access, unlike other systems that rely on cloud computing. This also provides an advantage in data security, as results can be produced without the need to share data with a cloud server provider.

It is also designed to be able to deliver results on a mobile device, eliminating the need to take a laptop into the field.

“You can build a more robust system that could be brought to resource-poor or less-developed regions, and it still works,” Schmidt said.

This improved system will work for any other biomarkers that Schmidt’s lab systems have used to detect in the past, such as biomarkers for COVID-19, Ebola, influenza, and cancer. Although they are currently focused on medical applications, the system could potentially be adapted for the detection of any type of signal.

To push the technology further, Schmidt and members of his lab plan to add even more dynamic signal processing capabilities to their devices. This will simplify the system and combine the processing techniques needed to detect signals at high and low concentrations of molecules. The team is also working to bring discrete parts of the setup into the integrated design of the optofluidic chip.