Fully Automated\nand AI-Assisted Optical Fiber Sensing\nSystem for Multiplexed and Continuous Brain Monitoring

Continuous and comprehensive brain monitoring is crucial\nfor timely\nidentification of changes or deterioration in brain function, enabling\nprompt intervention and personalized treatments. However, existing\nbrain monitoring systems struggle to offer continuous and accurate\nmonitoring of multiple brain biomarkers simultaneously. This study\nintroduces a multiplexed optical fiber sensing system for continuous\nand simultaneous monitoring of six cerebrospinal fluid (CSF) biomarkers\nusing tip-functionalized optical fibers and computational algorithms.\nOptimized machine learning models are developed and integrated for\nreal-time spectra analysis, allowing for precise and continuous readout\nof biomarker concentrations. The developed machine learning-assisted\nfiber optic sensing system exhibits high sensitivity (0.04, 0.38,\n0.67, 2.62, 0.0064, 0.33 <i>I</i>/<i>I</i><sub>0</sub> change per units of temperature, dissolved oxygen, glucose,\npH, Na<sup>+</sup>, Ca<sup>2+</sup>, respectively), reversibility,\nand selectivity toward target biomarkers with a total diameter less\nthan 2.5 mm. By monitoring brain metabolic and ionic dynamics, this\nsystem accurately identified brain physiology deterioration and recovery\nusing <i>ex vivo</i> traumatic brain injury models. Additionally,\nthe system successfully tracked biomarker fluctuations in clinical\nCSF samples with high accuracy (<i>R</i><sup>2</sup> >\n0.93),\ndemonstrating excellent sensitivity and selectivity in reflecting\ndisease progression in real time. These findings underscore the enormous\npotential of automated and multiplexed optical fiber sensing systems\nfor intraoperative and postoperative monitoring of brain physiologies.