AI-based differential diagnosis of dementia etiologies on multimodal data.

Differential diagnosis of dementia remains a challenge in neurology due to symptom overlap across etiologies, yet it is crucial for formulating early, personalized management strategies. Here, we present an artificial intelligence (AI) model that harnesses a broad array of data, including demographics, individual and family medical history, medication use, neuropsychological assessments, functional evaluations and multimodal neuroimaging, to identify the etiologies contributing to dementia in individuals. The study, drawing on 51,269 participants across 9 independent, geographically diverse datasets, facilitated the identification of 10 distinct dementia etiologies. It aligns diagnoses with similar management strategies, ensuring robust predictions even with incomplete data. Our model achieved a microaveraged area under the receiver operating characteristic curve (AUROC) of 0.94 in classifying individuals with normal cognition, mild cognitive impairment and dementia. Also, the microaveraged AUROC was 0.96 in differentiating the dementia etiologies. Our model demonstrated proficiency in addressing mixed dementia cases, with a mean AUROC of 0.78 for two co-occurring pathologies. In a randomly selected subset of 100 cases, the AUROC of neurologist assessments augmented by our AI model exceeded neurologist-only evaluations by 26.25%. Furthermore, our model predictions aligned with biomarker evidence and its associations with different proteinopathies were substantiated through postmortem findings. Our framework has the potential to be integrated as a screening tool for dementia in clinical settings and drug trials. Further prospective studies are needed to confirm its ability to improve patient care.

Improving Clinician Performance in Classifying EEG Patterns on the Ictal-Interictal Injury Continuum Using Interpretable Machine Learning.

BACKGROUND: In intensive care units (ICUs), critically ill patients are monitored with electroencephalography (EEG) to prevent serious brain injury. EEG monitoring is constrained by clinician availability, and EEG interpretation can be subjective and prone to interobserver variability. Automated deep-learning systems for EEG could reduce human bias and accelerate the diagnostic process. However, existing uninterpretable (black-box) deep-learning models are untrustworthy, difficult to troubleshoot, and lack accountability in real-world applications, leading to a lack of both trust and adoption by clinicians. METHODS: We developed an interpretable deep-learning system that accurately classifies six patterns of potentially harmful EEG activity - seizure, lateralized periodic discharges (LPDs), generalized periodic discharges (GPDs), lateralized rhythmic delta activity (LRDA), generalized rhythmic delta activity (GRDA), and other patterns - while providing faithful case-based explanations of its predictions. The model was trained on 50,697 total 50-second continuous EEG samples collected from 2711 patients in the ICU between July 2006 and March 2020 at Massachusetts General Hospital. EEG samples were labeled as one of the six EEG patterns by 124 domain experts and trained annotators. To evaluate the model, we asked eight medical professionals with relevant backgrounds to classify 100 EEG samples into the six pattern categories - once with and once without artificial intelligence (AI) assistance - and we assessed the assistive power of this interpretable system by comparing the diagnostic accuracy of the two methods. The model's discriminatory performance was evaluated with area under the receiver-operating characteristic curve (AUROC) and area under the precision-recall curve. The model's interpretability was measured with task-specific neighborhood agreement statistics that interrogated the similarities of samples and features. In a separate analysis, the latent space of the neural network was visualized by using dimension reduction techniques to examine whether the ictal-interictal injury continuum hypothesis, which asserts that seizures and seizure-like patterns of brain activity lie along a spectrum, is supported by data. RESULTS: The performance of all users significantly improved when provided with AI assistance. Mean user diagnostic accuracy improved from 47 to 71% (P<0.04). The model achieved AUROCs of 0.87, 0.93, 0.96, 0.92, 0.93, and 0.80 for the classes seizure, LPD, GPD, LRDA, GRDA, and other patterns, respectively. This performance was significantly higher than that of a corresponding uninterpretable black-box model (with P<0.0001). Videos traversing the ictal-interictal injury manifold from dimension reduction (a two-dimensional representation of the original high-dimensional feature space) give insight into the layout of EEG patterns within the network's latent space and illuminate relationships between EEG patterns that were previously hypothesized but had not yet been shown explicitly. These results indicate that the ictal-interictal injury continuum hypothesis is supported by data. CONCLUSIONS: Users showed significant pattern classification accuracy improvement with the assistance of this interpretable deep-learning model. The interpretable design facilitates effective human-AI collaboration; this system may improve diagnosis and patient care in clinical settings. The model may also provide a better understanding of how EEG patterns relate to each other along the ictal-interictal injury continuum. (Funded by the National Science Foundation, National Institutes of Health, and others.).

Trends in management of patients with new-onset refractory status epilepticus (NORSE) from 2016 to 2023: An interim analysis.

In response to the evolving treatment landscape for new-onset refractory status epilepticus (NORSE) and the publication of consensus recommendations in 2022, we conducted a comparative analysis of NORSE management over time. Seventy-seven patients were enrolled by 32 centers, from July 2016 to August 2023, in the NORSE/FIRES biorepository at Yale. Immunotherapy was administered to 88% of patients after a median of 3 days, with 52% receiving second-line immunotherapy after a median of 12 days (anakinra 29%, rituximab 25%, and tocilizumab 19%). There was an increase in the use of second-line immunotherapies (odds ratio [OR] = 1.4, 95% CI = 1.1-1.8) and ketogenic diet (OR = 1.8, 95% CI = 1.3-2.6) over time. Specifically, patients from 2022 to 2023 more frequently received second-line immunotherapy (69% vs 40%; OR = 3.3; 95% CI = 1.3-8.9)-particularly anakinra (50% vs 13%; OR = 6.5; 95% CI = 2.3-21.0), and the ketogenic diet (OR = 6.8; 95% CI = 2.5-20.1)-than those before 2022. Among the 27 patients who received anakinra and/or tocilizumab, earlier administration after status epilepticus onset correlated with a shorter duration of status epilepticus (ρ = .519, p = .005). Our findings indicate an evolution in NORSE management, emphasizing the increasing use of second-line immunotherapies and the ketogenic diet. Future research will clarify the impact of these treatments and their timing on patient outcomes.

Patient Handoff Practices at the Epilepsy Centers in the United States: A Survey of the Medical Directors.

PURPOSE: Communication failure is one of the most significant causes of medical errors. Providing care to patients with seizures at comprehensive epilepsy centers requires uninterrupted coverage and a multidisciplinary approach. However, handoff practices in these settings have not been comprehensively assessed, and recommendations for their standardization are currently lacking. The aim of this observational study was to define the scope of existing practices for patient handoffs across epilepsy centers in the United States and provide relevant recommendations. METHODS: A 79-question survey was developed to establish the patterns of transition of care for patients undergoing continuous EEG recording, including the periodicity of handoffs and specifics of the relevant workflow. With permission from the National Association of Epilepsy Centers (NAEC), the survey was distributed to the medical directors of all Level 3 and 4 NAEC-accredited epilepsy centers in the United States. RESULTS: The responses were obtained from 70 institutions yielding a survey response rate of 26%. Of these, more than 77% had established weekly handoff processes for both the epilepsy monitoring unit and continuous EEG (cEEG) monitoring services. However, only 53% and 43% of centers had procedures for daily service transfers for the patients admitted to the epilepsy monitoring unit or the patients undergoing cEEG, respectively. The patterns of handoffs were complex and utilized group handoffs in < 50% of institutions. In most centers (>70%), patient data transmitted through handoffs included history, clinical information, and EEG findings. However, templates were not applied to standardize this information. All participants agreed or strongly agreed that a culture of patient safety was maintained in their place of practice; however, 12% of participants felt that insufficient time was allowed to discuss these patients or carry out the handoffs without interruptions. CONCLUSIONS: Existing handoff practices are not uniform or fully established across epilepsy centers in the United States. This study recommends that guidelines for formal handoff procedures be developed and introduced as a quality metric for all NAEC-accredited epilepsy centers.

Memory loss and aberrant neurogenesis in mice exposed to patient anti-N-methyl-d-aspartate receptor antibodies.

OBJECTIVE: Anti-N-methyl-d-aspartate receptor (anti-NMDAR) encephalitis results in chronic epilepsy and permanent cognitive impairment. One of the possible causes of cognitive impairment in anti-NMDAR could be aberrant neurogenesis, an established contributor to memory loss in idiopathic drug-resistant epilepsy. We developed a mouse model of anti-NMDAR encephalitis and showed that mice exposed to patient anti-NMDAR antibodies for 2 weeks developed seizures and memory loss. In the present study, we assessed the delayed effects of patient-derived antibodies on cognitive phenotype and examined the corresponding changes in hippocampal neurogenesis. METHODS: Monoclonal anti-NMDAR antibodies or control antibodies were continuously infused into the lateral ventricle of male C56BL/6J mice (8-12 weeks) via osmotic minipumps for 2 weeks. The motor and anxiety phenotypes were assessed using the open field paradigm, and hippocampal memory and learning were assessed using the object location, Y maze, and Barnes maze paradigms during weeks 1 and 3-4 of antibody washout. The numbers of newly matured granule neurons (Prox-1+) and immature progenitor cells (DCX+) as well as their spatial distribution within the hippocampus were assessed at these time points. Bromodeoxyuridine (BrdU, 50 mg/kg, i.p., daily) was injected on days 2-12 of the infusion, and proliferating cell immunoreactivity was compared in antibody-treated mice and control mice during week 4 of the washout. RESULTS: Mice infused with anti-NMDAR antibodies demonstrated spatial memory impairment during week 1 of antibody washout (p = 0.02, t-test; n = 9-11). Histological analysis of hippocampal sections from these mice revealed an increased ectopic displacement of Prox-1+ cells in the dentate hilus compared to the control-antibody-treated mice (p = 0.01; t-test). Mice exposed to anti-NMDAR antibodies also had an impairment of spatial memory and learning during weeks 3-4 of antibody washout (object location: p = 0.009; t-test; Y maze: p = 0.006, t-test; Barnes maze: p = 0.008, ANOVA; n = 8-10). These mice showed increased ratios of the low proliferating (bright) to fast proliferating (faint) BrdU+ cell counts and decreased number of DCX+ cells in the hippocampal dentate gyrus (p = 0.006 and p = 0.04, respectively; t-tests) suggesting ectopic migration and delayed cell proliferation. SIGNIFICANCE: These findings suggest that memory and learning impairments induced by patient anti-NMDAR antibodies are sustained upon removal of antibodies and are accompanied by aberrant hippocampal neurogenesis. Interventions directed at the manipulation of neuronal plasticity in patients with encephalitis and cognitive loss may be protective and therapeutically relevant.

Comparative analysis of patients with new onset refractory status epilepticus preceded by fever (febrile infection-related epilepsy syndrome) versus without prior fever: An interim analysis.

Febrile infection-related epilepsy syndrome (FIRES) is a subset of new onset refractory status epilepticus (NORSE) that involves a febrile infection prior to the onset of the refractory status epilepticus. It is unclear whether FIRES and non-FIRES NORSE are distinct conditions. Here, we compare 34 patients with FIRES to 30 patients with non-FIRES NORSE for demographics, clinical features, neuroimaging, and outcomes. Because patients with FIRES were younger than patients with non-FIRES NORSE (median = 28 vs. 48 years old, p = .048) and more likely cryptogenic (odds ratio = 6.89), we next ran a regression analysis using age or etiology as a covariate. Respiratory and gastrointestinal prodromes occurred more frequently in FIRES patients, but no difference was found for non-infection-related prodromes. Status epilepticus subtype, cerebrospinal fluid (CSF) and magnetic resonance imaging findings, and outcomes were similar. However, FIRES cases were more frequently cryptogenic; had higher CSF interleukin 6, CSF macrophage inflammatory protein-1 alpha (MIP-1a), and serum chemokine ligand 2 (CCL2) levels; and received more antiseizure medications and immunotherapy. After controlling for age or etiology, no differences were observed in presenting symptoms and signs or inflammatory biomarkers, suggesting that FIRES and non-FIRES NORSE are very similar conditions.

AI-based differential diagnosis of dementia etiologies on multimodal data.

Differential diagnosis of dementia remains a challenge in neurology due to symptom overlap across etiologies, yet it is crucial for formulating early, personalized management strategies. Here, we present an AI model that harnesses a broad array of data, including demographics, individual and family medical history, medication use, neuropsychological assessments, functional evaluations, and multimodal neuroimaging, to identify the etiologies contributing to dementia in individuals. The study, drawing on 51,269 participants across 9 independent, geographically diverse datasets, facilitated the identification of 10 distinct dementia etiologies. It aligns diagnoses with similar management strategies, ensuring robust predictions even with incomplete data. Our model achieved a micro-averaged area under the receiver operating characteristic curve (AUROC) of 0.94 in classifying individuals with normal cognition, mild cognitive impairment and dementia. Also, the micro-averaged AUROC was 0.96 in differentiating the dementia etiologies. Our model demonstrated proficiency in addressing mixed dementia cases, with a mean AUROC of 0.78 for two co-occurring pathologies. In a randomly selected subset of 100 cases, the AUROC of neurologist assessments augmented by our AI model exceeded neurologist-only evaluations by 26.25%. Furthermore, our model predictions aligned with biomarker evidence and its associations with different proteinopathies were substantiated through postmortem findings. Our framework has the potential to be integrated as a screening tool for dementia in various clinical settings and drug trials, with promising implications for person-level management.

MyD88-mediated signaling is critical for the generation of seizure responses and cognitive impairment in a model of anti-N-methyl-D-aspartate receptor encephalitis.

OBJECTIVE: We previously demonstrated that interleukin-1 receptor-mediated immune activation contributes to seizure severity and memory loss in anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis. In the present study, we assessed the role of the myeloid differentiation primary response gene 88 (MyD88), an adaptor protein in Toll-like receptor signaling, in the key phenotypic characteristics of anti-NMDAR encephalitis. METHODS: Monoclonal anti-NMDAR antibodies or control antibodies were infused into the lateral ventricle of MyD88 knockout mice (MyD88-/-) and control C56BL/6J mice (wild type [WT]) via osmotic minipumps for 2 weeks. Seizure responses were measured by electroencephalography. Upon completion of the infusion, the motor, anxiety, and memory functions of the mice were assessed. Astrocytic (glial fibrillary acidic protein [GFAP]) and microglial (ionized calcium-binding adaptor molecule 1 [Iba-1]) activation and transcriptional activation for the principal inflammatory mediators involved in seizures were determined using immunohistochemistry and quantitative real-time polymerase chain reaction, respectively. RESULTS: As shown before, 80% of WT mice infused with anti-NMDAR antibodies (n = 10) developed seizures (median = 11, interquartile range [IQR] = 3-25 in 2 weeks). In contrast, only three of 14 MyD88-/- mice (21.4%) had seizures (0, IQR = 0-.25, p = .01). The WT mice treated with antibodies also developed memory loss in the novel object recognition test, whereas such memory deficits were not apparent in MyD88-/- mice treated with anti-NMDAR antibodies (p = .03) or control antibodies (p = .04). Furthermore, in contrast to the WT mice exposed to anti-NMDAR antibodies, the MyD88-/- mice had a significantly lower induction of chemokine (C-C motif) ligand 2 (CCL2) in the hippocampus (p = .0001, Sidak tests). There were no significant changes in the expression of GFAP and Iba-1 in the MyD88-/- mice treated with anti-NMDAR or control antibodies. SIGNIFICANCE: These findings suggest that MyD88-mediated signaling contributes to the seizure and memory phenotype in anti-NMDAR encephalitis and that CCL2 activation may participate in the expression of these features. The removal of MyD88 inflammation may be protective and therapeutically relevant.