Home AI News Enhancing Biomarker Detection with Machine Learning for Real-Time Diagnostic Devices

Enhancing Biomarker Detection with Machine Learning for Real-Time Diagnostic Devices

Enhancing Biomarker Detection with Machine Learning for Real-Time Diagnostic Devices

Sophisticated Systems for Biomarker Detection

UC Santa Cruz’s distinguished professor of electrical and computer engineering, Holger Schmidt, and his team have developed optofluidic chips to detect biomarkers. These molecules, such as DNA or proteins, are essential for diagnosing diseases in real-time and monitoring their progression.

To enhance their systems’ accuracy in classifying biomarkers, Schmidt’s graduate student, Vahid Ganjalizadeh, used machine learning. The deep neural network he created can classify particle signals with a remarkable accuracy of 99.8% in real-time. Moreover, this system is relatively affordable and portable, making it ideal for point-of-care applications.

Addressing Weak Signals

In field settings like health clinics, biomarker detectors may receive lower quality signals compared to labs. This can be due to various factors, including the use of cheaper chips to reduce costs or environmental conditions like temperature and humidity.

To tackle this challenge, Schmidt’s team developed a deep neural network with high-confidence weak signal identification capabilities. By training the neural network with known signals, it can recognize patterns and identify new signals with exceptional accuracy.

Efficient Signal Processing

The system employs a parallel cluster wavelet analysis (PCWA) approach to detect signal presence. Then, the neural network processes the potentially weak or noisy signal, identifying its source. Thanks to its real-time capabilities, users can receive results within seconds.

“Our focus is on maximizing the utilization of potentially low-quality signals, quickly and efficiently,” Schmidt stated.

Portable and Secure

A compact version of the neural network model runs on portable devices, such as the Google Coral Dev board. This not only reduces power consumption during processing but also allows for easier execution of artificial intelligence algorithms.

Furthermore, the system operates solely locally, eliminating the need for internet access. This provides a data security advantage as results can be produced without sharing sensitive data with cloud servers.

The system is also optimized for mobile devices, eliminating the necessity of carrying a laptop in the field.

Wide Applications

This improved system works for various biomarkers, including COVID-19, Ebola, flu, and cancer biomarkers. While the focus is currently on medical applications, the system has the potential to adapt to detect any type of signal.

Future Developments

Schmidt’s team plans to enhance their devices by integrating more dynamic signal processing capabilities. By simplifying the system, they aim to detect signals at both low and high molecule concentrations. They also intend to integrate discrete parts of the setup into the optofluidic chip’s design.

Source link


Please enter your comment!
Please enter your name here