There is an unmet need to differentiate NMOSD from MS due to the potential effects and safety risks associated with executing the wrong treatment regimen. Infrared spectroscopy coupled with a machine learning algorithm may be a promising point-of-care diagnostic tool.
Establishing a diagnosis of neuromyelitis optica spectrum disorder (NMOSD) differentiated from multiple sclerosis (MS) is critical to prevent complications. Although they have similar clinical characteristics, these disorders require the appropriate use of suitable treatment regimens.
Diagnostic criteria often involve the detection of water channel aquaporin-4 (AQP-4) or myelin oligodendrocyte glycoprotein (MOG) antibodies in the patient’s serum, although patients with NMOSD can be negative for these antibodies. Studies have found that combining Fourier-transform infrared (FTIR) spectroscopy with a machine learning algorithm can be a rapid, reliable way to distinguish NMOSD from MS.
Research investigators assessed the use of FTIR spectroscopy linked to a random forest classifier, which is a machine learning algorithm. They built the model using sera samples from NMOSD, relapse-remitting multiple sclerosis (RRMS), and healthy control subjects. Serum samples were gathered from patients with confirmed NMOSD in the NOMADMUS French database and RRMS patients from the OFSEP cohort, among others.
Results from the study found that patients with NMOSD were distinguished from healthy patients or patients with MS with a specificity and sensitivity of 100% each. No measurable confusion between the study groups was identified while using the machine learning algorithm. FTIR spectroscopy in combination with a random forest classifier can be a fast and cost-effective solution for differentiating NMOSD from RRMS, without regard to serostatus.
As a point-of-care diagnostic tool, FTIR spectroscopy with a forest classification machine learning algorithm can be easily implemented without the need for expertise in infrared spectroscopy. Results from this diagnostic tool can be obtained within minutes [1].
Source:
[1] El Khoury, Y., Gebelin, M., De Sèze, J., Patte-Mensah, C., Marcou, G., Varnek, A., Mensah-Nyagan, A., Hellwig, P., & Collongues, N. (2022). Rapid discrimination of neuromyelitis optica spectrum disorder and multiple sclerosis using machine learning on infrared spectra of sera. International Journal of Molecular Sciences, 23(5), 2791. https://doi.org/10.3390/ijms23052791
