# Prediction model for severe autoimmune encephalitis: a tool for risk assessment and individualized treatment guidance

## TL;DR

This paper introduces a new prediction model for severe autoimmune encephalitis to help doctors assess risk and tailor treatments.

## Contribution

A novel prediction model for severe autoimmune encephalitis based on clinical and laboratory data is developed and validated.

## Key findings

- The model includes variables like age, seizures, and intrathecal IgG synthesis rate.
- The model shows excellent discriminative capacity and calibration in validation tests.
- It is feasible for clinical use and can guide individualized treatment.

## Abstract

Severe autoimmune encephalitis (AE) can cause significant neurological deficits, status epilepticus, status dystonicus, and even death, which can be life-threatening to patients. Accurate risk stratification for severe AE progression is critical for optimizing therapeutic strategies. The comprehensive prediction models for severe AE based on routine clinical data and laboratory indicators remain lacking.

To develop and validate a prediction model for severe AE to optimize individualized treatment.

We collected clinical data and laboratory examination results from 207 patients with confirmed AE. The study population was divided into development and validation cohort. A prediction model for severe AE was constructed using a nomogram and was rigorously validated both internally and externally. Severe AE was defined as modified Rankin Scale (mRS) > 2 and Clinical Assessment Scale for Encephalitis (CASE) > 4.

The variables ultimately included in the nomogram for the severe AE predictive model were age, psychiatric and/or behavioral abnormalities, seizures, decreased level of consciousness, cognitive impairment, involuntary movements, autonomic dysfunction, and increased intrathecal IgG synthesis rate. It demonstrated excellent discriminative capacity and calibration through internal-external validation.

The prediction model has highly feasibility in clinical practice, and holds promise as an important tool for risk assessment and guiding individualized treatment in patients with AE.