188: Proteomics + Machine Learning for Lyme Neuroborreliosis Diagnosis

️ Episode 188: Proteomics + Machine Learning for Lyme Neuroborreliosis Diagnosis

In this episode of PaperCast Base by Base, we explore how large‑scale mass‑spectrometry proteomics of cerebrospinal fluid and plasma, paired with supervised machine learning, can distinguish Lyme neuroborreliosis from viral meningitis and non‑LNB controls in adults.

Study Highlights:
The authors analyzed 308 CSF and 207 plasma samples across development and validation cohorts to define host‑response protein signatures and train diagnostic classifiers. CSF proteomics yielded strong discrimination of LNB against viral meningitis and against controls, with independent‑cohort AUCs around 0.92 and 0.90, respectively, and highlighted immunoglobulin chains, complement factors, innate immune proteins, and cytoskeletal markers as key features. A plasma‑based model distinguishing LNB from controls achieved an AUC of about 0.80 in validation and captured systemic innate immunity, complement activation, lipid transport, and coagulation signatures. Across matrices, overlapping proteins illuminated compartmentalized immunity, with many immunoglobulins increased in CSF but relatively lower in plasma for LNB, and SHAP analyses prioritized features linked to humoral and innate responses as well as cell damage and migration.

Conclusion:
Machine‑learning‑assisted proteomics shows promise for less‑invasive diagnosis and monitoring of Lyme neuroborreliosis and could reduce reliance on lumbar puncture if validated prospectively.

Reference:
Nielsen AB, Fjordside L, Drici L, Ottenheijm ME, Rasmussen C, Henningsson AJ, Harritshøj LH, Mann M, Mens H, Lebech A‑M, Wewer Albrechtsen NJ. The diagnostic potential of proteomics and machine learning in Lyme neuroborreliosis. Nature Communications. 2025;16:9322. https://doi.org/10.1038/s41467-025-64903-z

License:
This episode is based on an open-access article published under the Creative Commons Attribution 4.0 International License (CC BY 4.0) – https://creativecommons.org/licenses/by/4.0/

Support:
If you'd like to support Base by Base, you can make a one-time or monthly donation here: https://basebybase.castos.com/