Machine learning model predicts prognosis in SARS-CoV-2-infected autoimmune encephalitis patients

Area of Science:

• Neurology
• Infectious Diseases
• Immunology

Background:

• Autoimmune encephalitis (AE) is a severe neurological condition.
• The impact of SARS-CoV-2 infection on AE prognosis is not fully understood.
• Developing predictive models for AE outcomes post-SARS-CoV-2 is crucial.

Purpose of the Study:

• To evaluate the neurological prognosis of patients with AE following SARS-CoV-2 infection.
• To develop and validate risk prediction models for worse AE prognosis.
• To identify clinical factors associated with AE outcomes after SARS-CoV-2.

Main Methods:

• Prospective, multicenter, observational cohort study.
• Multivariate logistic regression to identify prognostic factors.
• Machine learning strategies for risk prediction model development.

Main Results:

• SARS-CoV-2 infection occurred in 241 of 308 AE patients.
• Worse neurological prognosis was observed in 12.4% of AE patients three months post-infection.
• Risk factors for poor prognosis included pre-infection mRS score, immunotherapy during infection, and SARS-CoV-2-related drowsiness and gastrointestinal symptoms.
• A machine learning model demonstrated high discrimination accuracy (0.96) for predicting prognosis.

Conclusions:

• SARS-CoV-2 infection can exacerbate neurological symptoms in patients with AE.
• Machine learning models are feasible for predicting AE prognoses using clinical data.
• Early identification of risk factors can guide management and improve patient outcomes.