UC San Diego

OBJECTIVE: Large language models such as ChatGPT have demonstrated significant potential in question-answering within ophthalmology, but there is a paucity of literature evaluating its ability to generate clinical assessments and discussions. The objectives of this study were to (1) assess the accuracy of assessment and plans generated by ChatGPT and (2) evaluate ophthalmologists abilities to distinguish between responses generated by clinicians versus ChatGPT. DESIGN: Cross-sectional mixed-methods study. SUBJECTS: Sixteen ophthalmologists from a single academic center, of which 10 were board-eligible and 6 were board-certified, were recruited to participate in this study. METHODS: Prompt engineering was used to ensure ChatGPT output discussions in the style of the ophthalmologist author of the Medical College of Wisconsin Ophthalmic Case Studies. Cases where ChatGPT accurately identified the primary diagnoses were included and then paired. Masked human-generated and ChatGPT-generated discussions were sent to participating ophthalmologists to identify the author of the discussions. Response confidence was assessed using a 5-point Likert scale score, and subjective feedback was manually reviewed. MAIN OUTCOME MEASURES: Accuracy of ophthalmologist identification of discussion author, as well as subjective perceptions of human-generated versus ChatGPT-generated discussions. RESULTS: Overall, ChatGPT correctly identified the primary diagnosis in 15 of 17 (88.2%) cases. Two cases were excluded from the paired comparison due to hallucinations or fabrications of nonuser-provided data. Ophthalmologists correctly identified the author in 77.9% ± 26.6% of the 13 included cases, with a mean Likert scale confidence rating of 3.6 ± 1.0. No significant differences in performance or confidence were found between board-certified and board-eligible ophthalmologists. Subjectively, ophthalmologists found that discussions written by ChatGPT tended to have more generic responses, irrelevant information, hallucinated more frequently, and had distinct syntactic patterns (all P < 0.01). CONCLUSIONS: Large language models have the potential to synthesize clinical data and generate ophthalmic discussions. While these findings have exciting implications for artificial intelligence-assisted health care delivery, more rigorous real-world evaluation of these models is necessary before clinical deployment. FINANCIAL DISCLOSURES: The author(s) have no proprietary or commercial interest in any materials discussed in this article.

Encephaloceles are considered rare with an approximate incidence of 1 in 35,000, and sphenoid encephaloceles are even more uncommon.2 Two types of sphenoid encephaloceles exist: medial perisellar encephaloceles, and lateral sphenoidal encephaloceles. Surgical correction of the lateral sphenoid recess encephalocele is achieved via one of two endoscopic approaches: extended sphenoidotomy or transpterygopalatine. Extended sphenoidotomy is preferred if the angle between the access door and lateral extension of bone defect is greater than 35°1. Otherwise, the transpterygopalatine approach is used. Intraoperative video demonstrating an extended sphenoidotomy approach to correcting a lateral recess sphenoidal encephalocele has not previously been published. Here we present a case of a 41-year-old female who presented with meningitis, a cerebrospinal fluid leak, and an incidental sphenoid mass. Brain MRI redemonstrated the mass in the sphenoid sinus consistent with an encephalocele occupying Sternbergs Canal. The patient consented to the procedure. The video demonstrates the skull base approach, encephalocele extraction, collagen inlay, and nasal septal bone and vascularized pedicled nasoseptal flap placement. Postoperative imaging confirmed the placement of the collagen inlay and nasal septal bone autograft. The patient recovered from surgery and was discharged on post-operative day 3 with no cerebrospinal fluid (CSF) leak recurrence. Postoperative follow up demonstrated viable nasoseptal graft without evidence of CSF leak. For patients with favorable anatomy, an extended sphenoidotomy approach to lateral sphenoid sinus encephalocele resection is a preferred alternative to the transpterygoid approach. This surgical video demonstrates the technique for managing lateral sphenoid sinus encephaloceles occupying Sternbergs canal, including endonasal approach, encephalocele resection and posterior sphenoid wall repair.

Background

Setting

Methods

Results

Conclusions

Insulin is an important regulator of whole-body glucose homeostasis. In insulin sensitive tissues such as muscle and adipose, insulin induces the translocation of glucose transporter 4 (GLUT4) to the cell membrane, thereby increasing glucose uptake. However, insulin also signals in tissues that are not generally associated with glucose homeostasis. In the human reproductive endocrine axis, hyperinsulinemia suppresses the secretion of gonadotropins from gonadotrope cells of the anterior pituitary, thereby linking insulin dysregulation to suboptimal reproductive health. In the mouse, gonadotropes express the insulin receptor which has the canonical signaling response of IRS, AKT, and mTOR activation. However, the functional outcomes of insulin action on gonadotropes are unclear. Here, we demonstrate through use of an optimized cell fractionation protocol that insulin stimulation of the LβT2 gonadotropic cell line results in the unexpected translocation of GLUT1 to the plasma membrane. Using our high purity fractionation protocol, we further demonstrate that though Akt signaling in response to insulin is intact, insulin-induced translocation of GLUT1 occurs independently of Akt activation in LβT2 cells.

Despite ART, we detected occasional microglia containing cell-associated HIV RNA and HIV DNA integrated into open regions of the hosts genome (∼0.005%) should be corrected to: Despite ART, we detected occasional microglia containing cell-associated HIV RNA and HIV DNA integrated into open regions of the hosts genome (∼0.5%).

BACKGROUND: Neurophysiological tools have yielded valuable insights into the pathophysiology and treatment of psychosis. However, studies using event-related potentials (ERPs) have primarily focused on mean scores and neglected the within-person variability of ERP scores. The neglect of within-person variability of ERPs in the search for biomarkers might have resulted in crucial differences related to psychosis being missed. In this registered report, we aimed to determine whether distinct patterns of intraindividual variability in ERP biomarkers would be observed in people with a lifetime psychosis diagnosis. METHODS: Publicly available data posted to the National Institute of Mental Health Data Archive for 1R01MH110434-01 was obtained for 162 patients with a lifetime history of psychosis and 178 never-psychotic (NP) participants. Participants completed tasks that measured the auditory mismatch negativity (MMN), P300, error-related negativity, and reward positivity. Multilevel location-scale models were used to determine whether patients showed greater intraindividual variability of ERP scores than NP participants. RESULTS: Contrary to predictions, the groups did not differ in within-person variability of MMN frequency, P300, or error-related negativity; patients showed less variability in MMN duration than NP participants. Exploratory analyses of a subset of patients with schizophrenia showed greater variability of MMN in this group than in the NP group. Greater severity of thought disorder and activation symptoms were associated with higher intraindividual MMN variability. CONCLUSIONS: Distinct patterns of intraindividual variability in the measured ERPs were not observed for the broad group of people with lifetime psychotic disorders. Exploratory analyses suggest that intraindividual differences in ERPs are more relevant to schizophrenia and certain symptom dimensions than to psychotic disorders broadly, but research is needed to confirm these exploratory findings.

Purpose

Methods

Results

Conclusion