cbflow (IN)

School: New Horizon Public School

My project focuses on improving brain age prediction using multimodal neuroimaging and machine learning algorithms. Brain age based on MRI is an emerging biomarker for neurodegenerative disease risk, as deviation from normal aging trajectories may indicate underlying pathology. However, most models rely solely on structural MRI.

My work incorporates arterial spin labeling MRI to measure cerebral blood flow, providing crucial functional information related to vascular health not visible on conventional scans. I engineered novel features from the MRI data and systematically evaluated diverse machine learning approaches for brain age regression using different combinations of structural and functional biomarkers.

Key findings demonstrate models integrating cerebral blood flow metrics significantly improve prediction accuracy compared to just using structural MRI. This multimodal approach enables subtle deviations from normal brain aging to be detected, allowing early identification of high-risk individuals before onset of cognitive decline.

If validated in broader clinical cohorts, my models could be translated to routine care for early neurodegenerative disease screening and informing preventative interventions to maintain cognitive function. This advances UN SDG 3 for good health and well-being.

My project methodology combines state-of-the-art neuroimaging, sophisticated feature engineering, and rigorous machine learning evaluation. This serves as a template for developing reliable and reproducible diagnostic biomarkers that embody responsible AI principles. I aim to raise awareness about leveraging technology ethically for social good.

Sustainable Development Goals:
My project contributes to UN SDG 3 for good health and well-being by developing enhanced brain age prediction models that can serve as biomarkers for early detection of neurodegenerative disease risk. Deviations from normal brain aging trajectories on MRI can indicate underlying pathology years before onset of symptoms. By incorporating multimodal neuroimaging and machine learning, my models can sensitively identify accelerated aging associated with heightened risk for dementia.

If validated clinically, this approach could be translated to routine screening and personalized interventions. Those exhibiting accelerated brain aging could be selected for preventative therapies to maintain cognitive function with advancing age. My goal is to shift the paradigm from late-stage disease diagnosis to pre-emptive care through early detection of abnormal aging patterns indicative of neurodegeneration risk. This promises substantial impact on reducing the burden of age-related cognitive decline.