Unifying Mitochondrial Epigenetic Signatures and Deep Learning-Driven Nanophotonic Sensing for Early Cardiovascular Risk Assessment in Aging Populations
Vikas Gurjar1,2, Pradyumna Kumar Mishra1,2; 1Division of Environmental Biotechnology, Genetics & Molecular Biology (EBGMB), ICMR-National Institute for Research in Environmental Health (NIREH), Bhopal, India; 2Faculty of Medical Research, Academy of Scientific and Innovative Research (AcSIR), Ghaziabad, India
Abstract
This study presents a cohesive, multidisciplinary project that integrates advanced nano-biosensing, molecular profiling, and artificial intelligence to improve early diagnosis and risk assessment of cardiovascular diseases (CVDs), particularly among aging populations. Central to our approach is a novel quadruple-hybrid nano-photonic assay designed to simultaneously detect three critical circulating biomarkers: cell-free mitochondrial DNA (ccf-mtDNA), NT-proBNP, and exosomes. This innovative assay employs four sensing components aminopyrene-tethered and PLL-functionalized carbon nanodots for ccf-mtDNA and ccf-NAs, alongside antibody-conjugated graphene nanodots for NT-proBNP and exosomes. The platform demonstrated exceptional sensitivity (LOD: 0.369 fg) and operational stability while eliminating the need for nucleic acid amplification. Combined with a long short-term memory deep learning model, the biosensor achieved an impressive 99.86% classification accuracy in distinguishing between CVD and non-CVD samples. We complemented this technological advancement with a cross-sectional pilot study investigating mitochondrial-associated epigenetic stress signatures in two age groups: younger adults (18-38 years, n = 154) and older adults (39-65 years, n = 105). Findings indicated that older participants exhibited increased mtDNA methylation, elevated mtDNA copy number, and heightened NT-proBNP levels, all of which were correlated with reduced telomerase activity and increased inflammatory responses. These molecular insights were seamlessly integrated with biosensor data and analyzed using four machine learning models, with Random Forest achieving the highest prediction accuracy (0.984) for CVD risk. Collectively, this project exemplifies a powerful integration of biosensing technology, molecular diagnostics, and AI-analytics, providing a non-invasive, scalable, and precise strategy for early detection of CVD and personalized risk assessment, particularly relevant for age-related vulnerabilities.
Speaker
Prof. Pradyumna Kumar Mishra
ICMR-National Institute for Research in Environmental Health (NIREH), Bhopal, India
India
Discussion
Ask question
