Serum Fingerprinting Array for Indication-Agnostic Disease Screening


Serum Fingerprinting Array for Indication-Agnostic Disease Screening



Serum biomarker diagnostics are widely used in medical decision-making, but in many cases, are not sensitive and/or specific enough to enable screening. To address the unmet need for liquid biopsy screening technologies, MSK investigators have developed a machine-learning-enabled nanosensor array that captures a “disease fingerprint” in patient serum. The array outperformed the best existing clinical screening test for ovarian cancer (Kim, Nat Biomed Eng, 2022) while simultaneously reducing the costs associated with such antibody-based diagnostics. Investigators are currently building a substantive proof-of-concept data set in additional indications to develop this technology into a liquid biopsy-type screening/triage platform for many disease indications.

Proof of Concept: Ovarian Cancer

Over 22,000 women are diagnosed with ovarian cancer, and 14,000 die each year. If detected at stage I, the five-year survival rate is over 90%, but survival is poor when detected at later stages (~40% for stage III and ~15% for stage IV). Despite this, detection of stage I/II high-grade serous ovarian cancer (HGSOC), the most common and deadly form of ovarian cancer, occurs in only 29% of patients. This prompted the investigators to develop a nanosensor/machine-learning-based technology to acquire a liquid biopsy-type disease “fingerprint” to screen for early-stage ovarian cancer. Moderately sized data sets of ~270 patients were used to train and validate machine learning models to differentiate ovarian cancer from other conditions. The sensor detected ovarian cancer from patient sera with an accuracy approaching 95%–significantly better than existing methods (longitudinal serum CA125 measurements and second line TVS).

Technology Details

Sensor arrays were designed using quantum-defect-functionalized carbon nanotubes. Patient serum can be applied to the nanotubes where a near-infrared fluorescence spectrometer will automatically collect the sensor response and an optimized machine learning algorithm provides a probability of the disease condition based on the data. The selectivity and sensitivity of this technology may be increasingly improved over time, even after release, due to the capabilities of machine learning processes.


This technology has the potential to significantly change patient care. This method can be readily and rapidly adapted to the detection of many conditions without the need for a known biomarker. The sensor technology can be developed as an inexpensive and rapid screening tool and will cost much less than the multiplexing of conventional markers from a CA125 blood tests.




PCT application PCT/US2022/013190 was filed in January 2022.


Kim et al., Detection of ovarian cancer via the spectral fingerprinting of quantum-defect-modified carbon nanotubes in serum by machine learning. Nature Biomedical Engineering. 2022 (link)


Daniel Heller, PhD, Head, Cancer Nanotechnology Laboratory, Molecular Pharmacology Program, Sloan Kettering Institute, MSK


James Delorme, Ph.D.

Licensing Manager
E-mail: [email protected]