Evaluation of ChatGPT-generated medical responses: A systematic review and meta-analysis.

OBJECTIVE: Large language models (LLMs) such as ChatGPT are increasingly explored in medical domains. However, the absence of standard guidelines for performance evaluation has led to methodological inconsistencies. This study aims to summarize the available evidence on evaluating ChatGPT's performance in answering medical questions and provide direction for future research. METHODS: An extensive literature search was conducted on June 15, 2023, across ten medical databases. The keyword used was "ChatGPT," without restrictions on publication type, language, or date. Studies evaluating ChatGPT's performance in answering medical questions were included. Exclusions comprised review articles, comments, patents, non-medical evaluations of ChatGPT, and preprint studies. Data was extracted on general study characteristics, question sources, conversation processes, assessment metrics, and performance of ChatGPT. An evaluation framework for LLM in medical inquiries was proposed by integrating insights from selected literature. This study is registered with PROSPERO, CRD42023456327. RESULTS: A total of 3520 articles were identified, of which 60 were reviewed and summarized in this paper and 17 were included in the meta-analysis. ChatGPT displayed an overall integrated accuracy of 56 % (95 % CI: 51 %-60 %, I2 = 87 %) in addressing medical queries. However, the studies varied in question resource, question-asking process, and evaluation metrics. As per our proposed evaluation framework, many studies failed to report methodological details, such as the date of inquiry, version of ChatGPT, and inter-rater consistency. CONCLUSION: This review reveals ChatGPT's potential in addressing medical inquiries, but the heterogeneity of the study design and insufficient reporting might affect the results' reliability. Our proposed evaluation framework provides insights for the future study design and transparent reporting of LLM in responding to medical questions.

Anti-GFAP Antibody-Associated Hypertrophic Pachymeningitis.

BACKGROUND: Autoimmune glial fibrillary acidic protein (GFAP) astrocytopathy is an inflammatory central nervous system (CNS) disorder that usually presents as steroid responsive encephalitis, meningitis, myelitis, or meningoencephalomyelitis. Hypertrophic pachymeningitis (HP) is an uncommon disorder that causes a localized or diffuse thickening of the dura mater. Depending on the etiology, HP can be idiopathic or secondary to a wide variety of other diseases. There are no reports of autoimmune GFAP astrocytopathy presenting as HP. METHODS: In this case report, we describe a rare case of pediatric HP possibly associated with anti-GFAP antibody. RESULTS: A 13-year-old previously healthy girl presented with headache for nearly 8 months with left-sided peripheral facial palsy and left-sided abductor nerve palsy in the second month of course. Magnetic resonance imaging (MRI) of the brain revealed contrast enhancement of hypertrophic dura mater. Anti-GFAPα antibodies were positive in serum and cerebrospinal fluid. The patient improved clinically after steroid treatment with partial resolution of abnormal intracranial MRI lesions. CONCLUSION: The present study suggests that HP may be one of the clinical phenotypes for autoimmune GFAP astrocytopathy or GFAP antibody is a biomarker for HP.

Phenotypic Spectrum of CASPR2 and LGI1 Antibodies Associated Neurological Disorders in Children.

Objectives: The clinical data of patients with double-positive for leucine-rich glioma-inactivated protein 1 (LGI1) and contactin-associated protein-like 2 (CASPR2) antibodies is limited, particularly for children. This study aimed to investigate and summarize the clinical features and long-term prognosis of children's LGI1 and CASPR2 antibodies related to neurological disorders. Methods: We collected the clinical data and prognosis of patients with dual positive antibodies of CASPR2 and LGI1, hospitalized in the Department of Neurology, Children's Hospital of Chongqing Medical University. Furthermore, we summarized the clinical phenotypes of this disorder in children by reviewing the published literature. Results: Two patients presenting with variable neurological symptoms including pain, hypertension, profuse sweating, irritability, and dyssomnia from Children's Hospital of Chongqing Medical University were enrolled in this study. Together with the two patients, we identified 17 children with dual CASPR2 and LGI1 antibodies, including 12 males and 5 females. At the onset, the median age was 4.1 years (range 1-16, interquartile range 2.5-13.5), with 9 children younger than 5 years and 6 adolescents. Of the 17 patients, 11 were diagnosed with Morvan syndrome, 4 with acquired neuromyotonia, 1 with Guillain-Barré syndrome, and 1 with Guillain-Barré syndrome combined with Morvan syndrome. Dysautonomia (14/17, 82.3%), pain (13/17, 76.4%), sleep disorders (13/17, 76.4%), encephalopathy (12/17, 70.5%), and weight loss (10/17, 58.8%) were the most frequently described symptoms overall. No tumors were identified. Of the 17 patients, 13 received immunotherapy comprising IVIG combination of IVMP during the acute symptomatic phase followed by oral prednisolone to maintain remission (n = 7), the combination of IVIG, IVMP, oral prednisolone and methotrexate (n = 1), the combination of IVIG, IVMP, and mycophenolate mofetil (n = 1), the combination of IVIG, IVMP, oral prednisolone, and rituximab (n = 1), IVIG only (n = 2), IVMP only (n = 1). Median modified Rankin Scale (mRS) scores in the acute phase were 3 (range 1-4) and improved gradually. Over the follow-up (median 8.6 months, range 1-36 months), 52.9% (9/17) of the patients recovered completely; one patient relapsed and showed immunotherapy-dependent. Conclusion: LGI1 and CASPR2 double-positive antibodies associated with the neurological diseases can occur in children of all ages and involve multiple nervous systems. Morvan syndrome is the most common phenotype of this disorder. The long-term outcomes are mostly favorable upon immunotherapy.

External Assessment of the Anti-N-Methyl-D-Aspartate Receptor Encephalitis One-Year Functional Status Score in Chinese Pediatric Patients.

Objective: to assess the performance of the Anti-N-Methyl-D-Aspartate Receptor encephalitis (NMDAR) One-Year Functional Status (NEOS) score in predicting one-year functional outcome in Chinese children with anti-NMDAR encephalitis. Methods: children with anti-NMDAR encephalitis at the Children's Hospital of Chongqing Medical University were retrospectively enrolled from January 2014 to December 2020. Patients were categorized into two groups based on the modified Rankin Scale (mRS) at one-year follow-up. Discrimination of the NEOS score was assessed by the area under curve (AUC) of the receiver operating characteristic curve. Calibration of the NEOS score was assessed by comparing predicted probabilities with observed probabilities using a calibration curve and the Hosmer-Lemeshow test. The clinical practicability of the NEOS score was evaluated by performing a decision curve analysis. Results: one hundred seventy-five children (101 females and 74 males) with anti-NMDAR encephalitis and a median age of 7.7 years were enrolled. Of those, 149 (85.1%) had a good outcome at 1 year (mRS ≤ 2), and the remaining 26 (14.9%) had a poor outcome (mRS > 2). Patients with a higher NEOS score had a significantly higher mRS at one-year follow-up [Spearman r = 0.3878, 95% confidence interval (CI): 0.2500-0.5103, P < 0.001]. The AUC of the NEOS score was 0.870 (95% CI: 0.801-0.938, P < 0.001). The observed probability and predicted probability showed moderate consistency in the calibration curve and the Hosmer-Lemeshow test (P = 0.912). The decision curve analysis showed that using the NEOS score to predict one-year outcomes could provide additional net benefit during clinical practice. Conclusions: the NEOS score is a potentially reliable model to predict the one-year functional outcome in Chinese children with anti-NMDAR encephalitis.

Virus reactivation after immunotherapy of anti-NMDAR encephalitis secondary to herpes simplex encephalitis: A case report.

Herpes simplex encephalitis is the most common cause of sporadic fatal encephalitis. More than half of patients with herpes simplex encephalitis will die and the vast majority of survivors have severe neurologic sequelae without effective antiviral therapy. Some people experience relapses such as secondary anti-N-methyl-d-aspartate receptor encephalitis despite appropriate antiviral treatment. It is rare that virus reactivation after immunotherapy of anti-N-methyl-d-aspartate receptor encephalitis after herpes simplex encephalitis. In the present study, we retrospectively review one patient who showed this rare situation.

Adenovirus Infection-associated Central Nervous System Disease in Children.

BACKGROUND: Adenovirus (Adv) is a frequent etiology of acute respiratory tract infections. Although rare, neurologic manifestations are known to occur during Adv infection. METHODS: We retrospectively analyzed clinical, laboratory, outcome and the relationship between clinical characteristics and viral detection results in the cerebrospinal fluid (CSF) in children with Adv-associated central nervous system (CNS) dysfunction. RESULTS TWENTYONE: (1.5%) cases had Adv-associated CNS manifestations. The median age was 1.4 years and 20 (95%) were less than 5 years of age. Six (28%) were male. The most frequently cited CNS symptoms were altered consciousness (100%) and seizure (14.3%). Fourteen cases (73.7%) had abnormal electroencephalogram examination and 6 cases (37.5%) had abnormal imaging. None of the patients had received cidofovir administration. Twenty children recovered without sequelae and 1 patient died of respiratory failure. Patients with positive Adv polymerase chain reaction (n = 11) presented lower onset age compared with that of patients with negative Adv polymerase chain reaction (n = 10) in the CSF. Clinical manifestation, laboratory findings, imaging studies and electroencephalogram showed no significant difference between the 2 groups. CONCLUSION: Adv is a rare cause of CNS disease in children, mainly causing altered consciousness. Adv was detected in more cases in the respiratory tract than the CSF, but the majority of patients had the virus detected in both. The lack of Adv in the CSF does not exclude CNS involvement. Furthermore, the viral detection results in the CSF do not seem useful as an indicator of the severity of CNS disease.

Clinical characteristics of children with Guillain-Barré syndrome and factors associated with disease severity.

Guillain-Barré syndrome (GBS) is the leading cause of pediatric acute flaccid paralysis. This study aimed to summarize the clinical features of children with GBS and to explore factors associated with the severity of weakness. One hundred and twenty-two children with GBS (73 males and 49 females) were retrospectively analysed. The median age (IQR) at diagnosis was 4.0 years (2.9-7.2 years), and 26.2% of patients were at the age of 2-3 years. Of the 122 cases, 71 (58.2%) had an antecedent infection, 58 (47.5%) had cranial nerve involvement, 36 (29.1%) had dysautonomia, 77 (63.1%) had sensory symptoms, 28 (23.0%) had difficulty in breathing of which 15 (12.3%) patients required mechanical ventilation, and 8 (6.6%) had normal tendon reflex or hyperreflexia. Cytoalbuminologic dissociation of the cerebrospinal fluid was observed in 97 cases (82.9%). Further, 120 patients underwent nerve conduction studies: 76 (63.3%) exhibited demyelinating features whereas 36 (30.0%) had axonal type of CBS. 70.2% of patients could walk independently at 12 weeks. Fourteen (11.5%) patients were classified into the mild group [GBS disability score (GBS-DS) < 3] and 108 (88.5%) were classified into the severe group (GBS-DS ≥ 3). The incidence of cranial involvement (P = 0.038) and decreased tendon reflexes (P = 0.048) were significantly different between the two groups. These findings suggested that cranial nerve involvement is associated with severe muscle weakness in children with GBS.

Clinical Features of Acute Rhabdomyolysis in 55 Pediatric Patients.

Objective: This study was designed to compare the clinical manifestations, laboratory tests, etiology, and prognosis of children with acute rhabdomyolysis (RM) at various ages. This study was designed to analyze the risk factors for acute kidney injury (AKI) in children with RM and to identify the role of neuromuscular and autoimmune disease in children with RM. Methods: Clinical data for 55 children with RM were collected and statistically analyzed. Patients were stratified to an infant group (G1) (age <1 year), preschool group (G2) (age 1-6 year), school-age group (G3) (age 7-11 year), and an adolescent group (G4) (age 12-16 year). Results: The top three clinical manifestations were dark urine (52.7%), myalgia (38.2%), and fever (23.8%). Patients in G1 had fever (71.4%), vomiting (77.8%), and urinalysis abnormalities (14.3%), without triad clinical manifestations. Fifty percent of patients in G4 group had myalgia; 70.8% had dark urine; 75% had abnormal urine tests. The most common cause in each age group was as follows: sepsis (57.1%) in G1; hereditary neuromuscular diseases (44.4%) in G2; immune diseases (40%) in G3; strenuous exercise (50%) in G4. Logistic regression analysis shown that AKI was not corelated with age, gender, or peak creatine phosphokinase. AKI was, however, associated with presence of an electrolyte disorder. Conclusion: The clinical manifestations and laboratory findings in infants with acute RM are not typical and need to be taken seriously. The presence of an electrolyte disorder is a risk factor for AKI in children with RM. The most common pathogenesis of RM varies among age groups. Congenital hereditary metabolic disease and immune diseases should not be ignored as a cause of RM in children.