An Explainable-AI approach for Diagnosis of COVID-19 using MALDI-ToF Mass Spectrometry

Seethi, Venkata Devesh Reddy; LaCasse, Zane; Chivte, Prajkta; Bland, Joshua; Kadkol, Shrihari S.; Gaillard, Elizabeth R.; Bharti, Pratool; Alhoori, Hamed

doi:10.48550/arxiv.2109.14099

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, modified and new bioinformatics tools are required for large proteins, and particularly large post-translationally modified proteins require mass spectra. These tools need to report peak maxima within a comparatively wide defined range and then calculate the peak area, before applying machine learning for pattern recognition [35,36]. In addition, when applying machine learning to a clinical diagnostic as opposed to a laboratory-based problem in mass spectrometry pattern recognition, a high degree of uncertainty has to be allowed for in the training sets [37].…”

Section: Peak Identification-quantification and Bioinformaticsmentioning

confidence: 99%

A How to Guide: Clinical Population Test Development and Authorization of MALDI-ToF Mass Spectrometry-Based Screening Tests for Viral Infections

Iles

Zmuidinaite

et al. 2022

Viruses

View full text Add to dashboard Cite

Applying MALDI-ToF mass spectrometry as a clinical diagnostic test for viruses is different from that of bacteria, fungi and other micro-organisms. This is because the systems biology of viral infections, the size and chemical nature of specific viral proteins and the mass spectrometry biophysics of how they are quantitated are fundamentally different. The analytical challenges to overcome when developing a clinical MALDI-ToF mass spectrometry tests for a virus, particularly human pathogenic enveloped viruses, are sample enrichment, virus envelope disruption, optimal matrix formulation, optimal MALDI ToF MS performance and optimal spectral data processing/bioinformatics. Primarily, the instrument operating settings have to be optimized to match the nature of the viral specific proteins, which are not compatible with setting established when testing for bacterial and many other micro-organisms. The capacity to be a viral infection clinical diagnostic instrument often stretches current mass spectrometers to their operational design limits. Finally, all the associated procedures, from sample collection to data analytics, for the technique have to meet the legal and operational requirement for often high-throughput clinical testing. Given the newness of the technology, clinical MALDI ToF mass spectrometry does not fit in with standard criteria applied by regulatory authorities whereby numeric outputs are compared directly to similar technology tests that have already been authorized for use. Thus, CLIA laboratory developed test (LDT) criteria have to be applied. This article details our experience of developing a SAR-CoV-2 MALDI-ToF MS test suitable for asymptomatic carrier infection population screening.

show abstract

Section: Peak Identification-quantification and Bioinformaticsmentioning

confidence: 99%