by UCLA Engineering Institute for Technology Advancement

Deep Learning-Enhanced Chemiluminescence Vertical Flow Assay for High-Sensitivity Cardiac Troponin I Testing. Credit: Ozcan Lab / UCLA

A research team at UCLA has developed a deep learning-powered chemiluminescence vertical flow assay (CL-VFA) that brings clinical laboratory-grade cardiac troponin I (cTnI) testing to a portable, cost-effective point-of-care platform.

Their work, published in the journal Small, demonstrates how the integration of chemiluminescence-based biosensing, high-sensitivity imaging through a portable reader, and AI-driven data analysis enables rapid, highly sensitive cTnI quantification for the detection of myocardial infarction (MI), also known as heart attack, in diverse clinical settings.

This technology holds the potential to democratize access to fast and reliable cardiac diagnostics, particularly in resource-limited environments where advanced laboratory infrastructure is lacking.

Cardiovascular diseases remain the leading global cause of death, responsible for more than 19 million annual fatalities. The early detection of MI is crucial for reducing mortality and improving patient outcomes.

However, current high-sensitivity cTnI assays rely on large, expensive laboratory analyzers that require trained personnel, restricting access to life-saving cardiac diagnostics—particularly in low-resource settings where rapid clinical decision-making is essential.

To address this challenge, UCLA researchers have developed a novel point-of-care diagnostics platform that offers high-sensitivity troponin testing in a compact, portable, and cost-effective design.

This research was led by Dr. Aydogan Ozcan, Chancellor's Professor of Electrical & Computer Engineering and the associate director of the California NanoSystems Institute (CNSI) at UCLA, in collaboration with Professor Dino Di Carlo of the UCLA Bioengineering Department, Professor Omai Garner, the director of UCLA Clinical Microbiology Lab.

The first authors of the paper are Dr. Gyeo-Re Han, a postdoctoral researcher, and Artem Goncharov, a graduate student at UCLA Electrical & Computer Engineering Department.

This platform features a robust integration of deep learning-driven computational analysis and highly sensitive chemiluminescence biosensing. This innovation allows for the detection of cTnI at levels as low as 0.1–0.2 pg/mL and an extensive dynamic range from less than 1 pg/mL to 100 ng/mL.

These specifications outperform existing point-of-care devices, effectively meeting the clinical standards for high-sensitivity troponin testing—an essential factor in early MI diagnosis and risk stratification. This point-of-care sensor requires only 50 µL of serum and features a streamlined workflow, potentially enabling medical staff to perform tests with simplicity. It provides cTnI results in just 25 min for rapid clinical decision-making.

"This technology represents a major step toward democratizing high-quality cardiac diagnostics," said Professor Ozcan. "By combining AI-powered analysis, chemiluminescence biosensing, and portable high-sensitivity imaging, we can bridge the gap between central laboratory testing and real-time clinical decision-making in emergency rooms, rural clinics, and decentralized health care settings."

This computational sensor operates in two key stages: an immunoassay phase followed by washing and a chemiluminescence signal generation phase. During the immunoassay phase, a polymerized enzyme-based conjugate binds to cTnI in the serum.

In the signal generation phase, a chemiluminescent material is activated through the enzymatic reaction, producing light signal captured by a custom-designed portable reader. A deep learning-driven algorithm then processes these images to infer cTnI concentrations in the measured serum sample.

"The ability to achieve such high sensitivity in a portable system is a game-changer for point-of-care troponin testing," said Dr. Gyeo-Re Han. "Our approach successfully combines robust CL biosensing with AI-driven data processing, making it both highly accurate and practical for real-world use."

The researchers rigorously validated their sensor platform using clinical serum samples. In a blinded validation study with patient samples, their sensor showed a strong correlation with an FDA-cleared laboratory analyzer, demonstrating its reliability, clinical accuracy, and potential for real-world diagnostic applications.

Beyond its high performance, this sensor is also designed for affordability. Traditional benchtop chemiluminescence analyzers cost more than ~$10,000–20,000. In contrast, the UCLA reader system, which is built on a custom optical imager, costs ~$222, while each test is priced at ~$4.

This cost-effectiveness of the sensor makes it an ideal solution for expanding access to cardiac diagnostics in primary care clinics, pharmacies, and mobile health units, particularly in resource-constrained settings.

"The affordability of our platform is key to making high-sensitivity cardiac testing truly accessible," said Professor Di Carlo. "By significantly reducing both equipment and per-test costs, we are bringing lab-quality diagnostics closer to the patients who need it most."

The researchers envision further expanding this paper-based sensor platform by integrating multiplexed detection of multiple cardiovascular biomarkers, enabling comprehensive cardiac risk assessment in a single test. The high sensitivity, portability, simplicity, and cost-effectiveness of this sensing platform establish it as a practical alternative to traditional laboratory-based testing, bringing high-sensitivity cardiac diagnostics closer to patients.

By democratizing access to reliable and rapid biomarker testing, this innovation has the potential to enhance clinical decision-making, improve patient outcomes, and expand cardiac care globally—especially in resource-limited and decentralized health care settings.

More information: Gyeo‐Re Han et al, Deep Learning‐Enhanced Chemiluminescence Vertical Flow Assay for High‐Sensitivity Cardiac Troponin I Testing, Small (2025). DOI: 10.1002/smll.202411585

Journal information: Small 

Provided by UCLA Engineering Institute for Technology Advancement