(299a) Invited Talk: Carbon Nanotube Quantum Well Defect Emission for Machine Learning-Guided Diagnostics

We employ the photoluminescence of single-walled carbon nanotube (SWCNTs), and covalent sp3 quantum well defects on SWCNTs, to develop new diagnostic methods for cancer and other diseases. Serum biomarker measurements are widely used for diagnosis, but these markers largely provide low sensitivity and specificity. We developed a method using defect-modified SWCNTs to identify a “disease fingerprint” through the collection of large data sets of molecular binding interactions to an array of quantum defect-modified carbon nanotubes. We found that a library of modified SWCNTs exhibited differentiated spectral variation in response to an ensemble of molecular binding events in patient serum. Via machine learning algorithms, we built a prediction model of nanosensor responses that reliably identified ovarian cancer substantially better than the established, FDA-approved biomarker, CA125. We have expanded this approach to other indications without known biomarkers, providing a general method to identify diseases.