UCLA researchers in the Department of Cardiology have developed a nanoengineered immune sensor that uses machine learning to detect multiple circulating biomarkers for rapid, accurate prediction of acute thrombosis.
BACKGROUND: Acute thrombosis represents a significant contributor to both acute and chronic illnesses. Moreover, the correlation between respiratory viral infections, such as COVID-19, and clinical thrombotic events is well established, emphasizing the importance of personalized prediction and prompt intervention. The rapid and precise stratification of thrombotic risk among hospitalized or high-risk patients with cardiometabolic diseases and/or cancers remains an unmet clinical need. Current biomarker assays, including D-dimer and sP-selectin, as well as traditional immunoassays, are limited by lengthy processing times, restricted multiplexing capabilities, and suboptimal sensitivity and specificity for early detection of thrombotic events. Consequently, there is a critical need for a point-of-care multiplex diagnostic platform capable of rapidly measuring multiple circulating biomarkers and using data-driven clustering analytics to identify patients at elevated risk of thrombosis before the onset of overt clinical symptoms.
INNOVATION: Researchers at UCLA have developed a nanoengineered multichannel immunosensor array for the simultaneous detection of multiple thrombosis-associated biomarkers, including C-reactive protein (CRP), calprotectin, soluble P-selectin (sP-selectin), and D-dimer. The sensor platform integrates carbon nanotube and nanofiber electrode channels, optimized antibody pairs, and an electrochemical readout to attain high sensitivity and rapid detection within minutes. The biomarker concentration data were subsequently analyzed using unsupervised machine learning clustering algorithms to stratify fifty-three patient specimens into thrombotic risk groups, achieving accurate predictions in forty-nine of fifty-three specimens. In summary, UCLA researchers have engineered a technology that combines rapid, multiplexed biosensing with unsupervised clustering to accurately predict thrombosis.
POTENTIAL APPLICATIONS:
- Point-of care stratification of thrombotic risk
- Monitoring of high-risk patients
- Can be used in conjunction with anti-thrombotic therapy
ADVANTAGES:
- Rapid turnaround time (minutes)
- Multiplex capacity
- Unsupervised analytics
- Validated on patient samples
- Flexible platform (can extend to include additional biomarkers or adapted for other pathologies)
DEVELOPMENT-TO-DATE: Research personnel at UCLA have pioneered a nanoengineered immunosensor platform, subsequently validating its efficacy in vitro with specimens obtained from human patients. The platform has exhibited high sensitivity and reproducibility. The assay underwent testing with 53 clinical plasma samples derived from patients, and unsupervised machine learning algorithms were employed to accurately stratify thrombotic risk in 49 of these 53 samples.
Related Papers (from the inventors only):
Wang K, Wang S, Margolis S, Cho JM, Zhu E, Dupuy A, Yin J, Park SK, Magyar CE, Adeyiga OB, Jensen KS, Belperio JA, Passam F, Zhao P, Hsiai TK. Rapid prediction of acute thrombosis via nanoengineered immunosensors with unsupervised clustering for multiple circulating biomarkers. Sci Adv. 2024 Dec 13;10(50):eadq6778. doi: 10.1126/sciadv.adq6778. Epub 2024 Dec 11. PMID: 39661669; PMCID: PMC11633740.
Keywords: Immunosensor, biosensor, multichannel electrode, carbon nanofiber sensor, point-of-care diagnostics, multiplex biomarker detection, thrombosis, unsupervised machine learning, disease detection, D-dimer, CRP, sP-selectin, hierarchical clustering, risk prediction, nanoengineering