Early autoimmunity and outcome in virus encephalitis: a retrospective study based on tissue-based assay.

To explore the autoimmune response and outcome in the central nervous system (CNS) at the onset of viral infection and correlation between autoantibodies and viruses. METHODS: A retrospective observational study was conducted in 121 patients (2016-2021) with a CNS viral infection confirmed via cerebrospinal fluid (CSF) next-generation sequencing (cohort A). Their clinical information was analysed and CSF samples were screened for autoantibodies against monkey cerebellum by tissue-based assay. In situ hybridisation was used to detect Epstein-Barr virus (EBV) in brain tissue of 8 patients with glial fibrillar acidic protein (GFAP)-IgG and nasopharyngeal carcinoma tissue of 2 patients with GFAP-IgG as control (cohort B). RESULTS: Among cohort A (male:female=79:42; median age: 42 (14-78) years old), 61 (50.4%) participants had detectable autoantibodies in CSF. Compared with other viruses, EBV increased the odds of having GFAP-IgG (OR 18.22, 95% CI 6.54 to 50.77, p<0.001). In cohort B, EBV was found in the brain tissue from two of eight (25.0%) patients with GFAP-IgG. Autoantibody-positive patients had a higher CSF protein level (median: 1126.00 (281.00-5352.00) vs 700.00 (76.70-2899.00), p<0.001), lower CSF chloride level (mean: 119.80±6.24 vs 122.84±5.26, p=0.005), lower ratios of CSF-glucose/serum-glucose (median: 0.50[0.13-0.94] vs 0.60[0.26-1.23], p=0.003), more meningitis (26/61 (42.6%) vs 12/60 (20.0%), p=0.007) and higher follow-up modified Rankin Scale scores (1 (0-6) vs 0 (0-3), p=0.037) compared with antibody-negative patients. A Kaplan-Meier analysis revealed that autoantibody-positive patients experienced significantly worse outcomes (p=0.031). CONCLUSIONS: Autoimmune responses are found at the onset of viral encephalitis. EBV in the CNS increases the risk for autoimmunity to GFAP.

Synthetic data accelerates the development of generalizable learning-based algorithms for X-ray image analysis

Artificial intelligence (AI) now enables automated interpretation of medical images. However, AI's potential use for interventional image analysis remains largely untapped. This is because the post hoc analysis of data collected during live procedures has fundamental and practical limitations, including ethical considerations, expense, scalability, data integrity and a lack of ground truth. Here we demonstrate that creating realistic simulated images from human models is a viable alternative and complement to large-scale in situ data collection. We show that training AI image analysis models on realistically synthesized data, combined with contemporary domain generalization techniques, results in machine learning models that on real data perform comparably to models trained on a precisely matched real data training set. We find that our model transfer paradigm for X-ray image analysis, which we refer to as SyntheX, can even outperform real-data-trained models due to the effectiveness of training on a larger dataset. SyntheX provides an opportunity to markedly accelerate the conception, design and evaluation of X-ray-based intelligent systems. In addition, SyntheX provides the opportunity to test novel instrumentation, design complementary surgical approaches, and envision novel techniques that improve outcomes, save time or mitigate human error, free from the ethical and practical considerations of live human data collection.Simulated data is an alternative to real data for medical applications where interventional data are needed to train AI-based systems. Gao and colleagues develop a model transfer paradigm to train deep networks on synthetic X-ray data and corresponding labels generated using simulation techniques from CT scans. The approach establishes synthetic data as a viable resource for developing machine learning models that apply to real clinical data.