Comparative Performance of Electrodiagnostic Criteria in Guillain-Barré Syndrome and the Impact of Machine-assisted Analysis
Jee Eun Oh1, Michael Skolka3, Grace Swart4, Elke Schipani1, Shahar Shelly5, Stefan Stålberg6, Benjamin Stålberg6, Abe Gardner6, Jacob Price6, Divyanshu Dubey1, Marcus Vinicius Pinto1, Reece Hass1, Jayawant Mandrekar2, Ruple Laughlin1, William Litchy1, Christopher Klein1
1Department of Neurology, 2Clinical Trials and Biostatistics, Mayo Clinic Rochester, 3Department of Neurology, Mayo Clinic Jacksonville, 4University of Sydney, Royal Prince Alfred Hospital, 5Department of Neurology and The Technion Institute of Technology, Rambam Medical Center, 6Cadwell Industries
Objective:
To evaluate the diagnostic performance of current nerve conduction criteria for Guillain-Barré syndrome (GBS) across disease stages and assess whether integrating a machine-assisted electrodiagnostic (EDX) calculator based on the 2021 EAN/PNS CIDP criteria can improve diagnostic specificity.
Background:
Accurate early diagnosis of GBS remains challenging, as electrophysiologic findings may be subtle early after symptom onset, and traditional criteria often trade sensitivity for specificity. Umapathi and VEGBS criteria improve early detection but can yield false positives among mimics. Incorporating machine-based EDX support tools may refine specificity and aid clinical decision-making.
Design/Methods:
We retrospectively analyzed nerve conductions from 24 GBS and 68 controls admitted to Mayo Clinic (Rochester, MN, and Jacksonville, FL). Brighton level 1–2 criteria were met in GBS patients with symptom onset in 2 cases <5 days, 6 cases <8 days, 14 cases <15 days, and 12 cases >15 days. All 68 controls had GBS within the initial differential (26 other neuropathies, 16 myopathy, 12 generalized weakness, 5 MG, and 9 other neuromuscular diagnoses). Sensitivity and specificity were calculated using Umapathi and VEGBS criteria and compared with a machine-assisted calculator applying 2021 EAN/PNS CIDP possible and definite demyelinating criteria.
Results:
Umapathi criteria showed the highest early sensitivity (100% < 8 days) but low specificity (25%), while VEGBS maintained high specificity (93%) but poor early sensitivity (33% < 8 days). In contrast, the machine-assisted EAN/PNS CIDP criteria achieved moderate sensitivity (83% < 8 days) with improved specificity (69–71%), offering a better balance for diagnostic accuracy as disease duration increased.
Conclusions:
Electrodiagnostic criteria are highly sensitive for GBS when applied early but lack specificity in isolation. Integrating machine-assisted tools applying 2021 CIDP criteria enhances specificity and supports clinical decision-making. Incorporating such tools into real-time EDX workflows may improve diagnostic accuracy and help distinguish GBS from mimics during acute presentations.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.