Related Experiment Video
Updated: Jan 9, 2026
Author Spotlight: Expanding the Scope of Multiplex Immunoassays for Lyme Borreliosis Diagnostics and Pathogen Research
Published on: July 14, 2023
Machine learning-supported framework for the classification of mpox infection and MVA immunization from multiplexed
Rebecca Surtees1, Fridolin Treindl2, Shakhnaz Akhmedova3
1Highly Pathogenic Viruses (ZBS 1), Centre for Biological Threats and Special Pathogens, German Consultant Laboratory for Poxviruses, WHO Collaborating Centre for Emerging Infections and Biological Threats, Robert Koch Institute, Berlin, Germany.
A new machine learning assay accurately distinguishes mpox virus infection from vaccination immunity. This tool is crucial for monitoring orthopoxvirus exposure and enhancing mpox serosurveillance efforts.
Area of Science:
- Infectious Diseases
- Immunology
- Computational Biology
Background:
- The 2022 global mpox outbreak emphasized the risk of zoonotic diseases and the need for precise serological tools.
- Distinguishing monkeypox virus (MPXV) infection from Modified Vaccinia Ankara (MVA) vaccination-induced immunity is challenging due to cross-reactive antibodies.
Purpose of the Study:
- To develop and validate a machine learning-assisted serological assay capable of differentiating MPXV infection from MVA vaccination and pre-immune sera.
- To assess the performance of various machine learning algorithms in classifying orthopoxvirus immune status.
Main Methods:
- A bead-based serological multiplex assay targeting antibody responses to 15 poxviral antigens was developed.
- Six machine learning algorithms were evaluated, with the Gradient Boosting Classifier (GBC) selected for its superior performance.
- The assay was validated on epidemiological and independent cohorts, including at-risk men who have sex with men (MSM).
Main Results:
- The Gradient Boosting Classifier (GBC) achieved the highest performance (F1 score = 0.83) in classifying sera from the 2022 mpox outbreak and an MSM cohort (n=1,260).
- In an independent validation cohort (n=143), GBC demonstrated robust detection of MPXV infections (F1 score = 0.70), including breakthrough cases, with 88% specificity and 92% sensitivity.
- The ML-assisted assay accurately classified orthopoxvirus immune status.
Conclusions:
- Integrating machine learning with high-dimensional serology provides an accurate method for classifying orthopoxvirus immune status.
- This approach offers a scalable framework for mpox serosurveillance and preparedness for future outbreaks.
- The developed assay effectively distinguishes between mpox infection and MVA vaccination.
