Assessment of a Large Language Model's Responses to Questions and Cases About Glaucoma and Retina Management.

Importance: Large language models (LLMs) are revolutionizing medical diagnosis and treatment, offering unprecedented accuracy and ease surpassing conventional search engines. Their integration into medical assistance programs will become pivotal for ophthalmologists as an adjunct for practicing evidence-based medicine. Therefore, the diagnostic and treatment accuracy of LLM-generated responses compared with fellowship-trained ophthalmologists can help assess their accuracy and validate their potential utility in ophthalmic subspecialties. Objective: To compare the diagnostic accuracy and comprehensiveness of responses from an LLM chatbot with those of fellowship-trained glaucoma and retina specialists on ophthalmological questions and real patient case management. Design, Setting, and Participants: This comparative cross-sectional study recruited 15 participants aged 31 to 67 years, including 12 attending physicians and 3 senior trainees, from eye clinics affiliated with the Department of Ophthalmology at Icahn School of Medicine at Mount Sinai, New York, New York. Glaucoma and retina questions (10 of each type) were randomly selected from the American Academy of Ophthalmology's commonly asked questions Ask an Ophthalmologist. Deidentified glaucoma and retinal cases (10 of each type) were randomly selected from ophthalmology patients seen at Icahn School of Medicine at Mount Sinai-affiliated clinics. The LLM used was GPT-4 (version dated May 12, 2023). Data were collected from June to August 2023. Main Outcomes and Measures: Responses were assessed via a Likert scale for medical accuracy and completeness. Statistical analysis involved the Mann-Whitney U test and the Kruskal-Wallis test, followed by pairwise comparison. Results: The combined question-case mean rank for accuracy was 506.2 for the LLM chatbot and 403.4 for glaucoma specialists (n = 831; Mann-Whitney U = 27976.5; P < .001), and the mean rank for completeness was 528.3 and 398.7, respectively (n = 828; Mann-Whitney U = 25218.5; P < .001). The mean rank for accuracy was 235.3 for the LLM chatbot and 216.1 for retina specialists (n = 440; Mann-Whitney U = 15518.0; P = .17), and the mean rank for completeness was 258.3 and 208.7, respectively (n = 439; Mann-Whitney U = 13123.5; P = .005). The Dunn test revealed a significant difference between all pairwise comparisons, except specialist vs trainee in rating chatbot completeness. The overall pairwise comparisons showed that both trainees and specialists rated the chatbot's accuracy and completeness more favorably than those of their specialist counterparts, with specialists noting a significant difference in the chatbot's accuracy (z = 3.23; P = .007) and completeness (z = 5.86; P < .001). Conclusions and Relevance: This study accentuates the comparative proficiency of LLM chatbots in diagnostic accuracy and completeness compared with fellowship-trained ophthalmologists in various clinical scenarios. The LLM chatbot outperformed glaucoma specialists and matched retina specialists in diagnostic and treatment accuracy, substantiating its role as a promising diagnostic adjunct in ophthalmology.

Ophthalmology and Social Media: An In-Depth Investigation of Ophthalmologic Content on Instagram.

Purpose: Social media has become a popular source of health information for patients. This study aimed to characterize the top-performing ophthalmologic posts on a large social media platform to better understand the spread of ophthalmic information via social media. Materials and Methods: This was a web-based study that searched for ophthalmology-related posts on Instagram, with subjects being users who posted ophthalmic content. A list of 36 ophthalmology-related hashtags, including the most common diagnoses and procedures identified from the IRIS Registry, was queried. For each hashtag, data were collected for "Top 9 posts" (as ranked by Instagram's engagement-based algorithm) at three different time points. Posts were analyzed for the poster's background, credentials, post format, content, caption length, and engagement level. Results: Of the top-performing posts analyzed (n = 972), the most frequent post format was a photo (82.2%), followed by video (8.8%) and graphic (8.4%). Ophthalmologists (35.8%) authored the highest number of posts, followed by patients (27.1%), optometrists (20.1%), and organizations (12.7%). The highest average engagement level ratios (ELRs) belonged to ophthalmologists-in-training (0.096), followed by patients (0.084), optometrists (0.070), all ophthalmologists (0.067) and organizations (0.051); p < 0.001. The most engaging type of content was self-promotional (0.118) and personal experience-related (0.091); educational content was the least engaging (0.059) even though it comprised the majority of posts (56%); (p < 0.001). Characteristics that predicted the highest ELRs (reaching 80th percentile) were captions and/or images that featured personal experiences (3.335 OR), whitecoats (3.259), and those authored by ophthalmologist trainees (3.172); (p < 0.01). The least engaging were those featuring fundus photos (0.281), educational content (0.359), and authored by organizations (0.428); (p < 0.05). Conclusion: The majority of ophthalmologic content on Instagram is authored by non-ophthalmologists, with educational content being the least engaging. Practicing ophthalmologists have an opportunity to reach more patients through social media by incorporating specific features known to drive post engagement and reach. Precis: Social media has become a popular source of health information for patients. Our study demonstrates that the majority of ophthalmology content on Instagram is authored by non- ophthalmologists, with educational content being the least engaging. Practicing ophthalmologists have an opportunity to reach a wider audience through social media by incorporating specific features known to drive post engagement and reach.

Serum molecular biomarkers in neuromyelitis optica and multiple sclerosis.

BACKGROUND: Neuromyelitis optica spectrum disorder (NMOSD) is a rare and severe inflammatory demyelinating disorder of the central nervous system (CNS), which mainly affects the optic nerves and spinal cord. The aims of this study were to determine whether the expression levels of serological cytokines could distinguish 1) NMOSD from healthy controls (HCs); and 2) NMOSD patients with and without the aquaporin-4 (AQP4) antibody biomarker from each other; and 3) NMOSD patients without the antibody to AQP4 from MS patients. METHODS: The expression levels of 200 proteins in serum from 41 NMOSD (32 with antibodies to AQP4, 9 without antibodies to AQP4), 12 MS patients, and 34 HCs were measured using glass-based antibody arrays. None of the patients received any immunosuppressive treatment. In parallel, the correlation between protein expression in NMOSD/MS patients and clinical traits was determined with Weighted Gene Co-expression Network Analysis (WGCNA). RESULTS: Thirty-nine serological proteins were differentially expressed in NMOSD patients compared to HCs, with 29 of these proteins not observed in MS patients. In addition, the data reveal 15 differentially-expression proteins (DEPs) between AQP4-IgG seronegative and AQP4-IgG seropositive NMOSD patients, and 9 DEPs between NMOSD and MS patients who did not have AQP4-IgG. CONCLUSION: Serological IL-17B is significantly upregulated in both NMOSD and MS patients compared to HCs, and could be a key biomarker of NMOSD and MS. Serological VEGF, MPIF-1 and NrCAM were positively associated with AQP4-IgG titer. We also demonstrate that EGF may be involved in the breakdown of the BBB by downregulating Claudin-5.

Optimizing the ocular surface prior to cataract surgery.

PURPOSE OF REVIEW: Ocular surface disease can significantly impact the outcomes of cataract surgery. Recent studies have examined the efficacy of several new dry eye disease (DED) therapies, the extent to which epithelial debridement affects keratometric measurements in epithelial basement membrane dystrophy (EBMD) and Salzmann nodular degeneration (SND), and the predictability of refractive error following combined pterygium and cataract removal. This review aims to incorporate these newer studies in updating and further emphasizing the need for careful management and optimization of common ocular surface conditions prior to cataract surgery. RECENT FINDINGS: Common ocular surface conditions such as DED, EBMD, SND, and pterygium can cause significant irregular astigmatism and higher-order aberrations. Their resolution can substantially alter biometry measurements in preparation for cataract surgery, affecting the final visual outcome. Newer therapies for DED, such as topical lifitegrast and thermal pulsation treatment, can aid in this optimization process. If superficial keratectomy or excisions of lesions on the ocular surface are performed, sufficient healing time is needed to allow the ocular surface to reach stability prior to biometry measurements. SUMMARY: Ocular surface optimization is key to successful cataract surgery planning and reaching desired outcomes.

Quantitative Detection of Anti-SARS-CoV-2 Antibodies Using Indirect ELISA.

OBJECTIVE: Real-time reverse transcription-polymerase chain reaction is the gold standard for the diagnosis of COVID-19, but it is necessary to utilize other tests to determine the burden of the disease and the spread of the outbreak such as IgG-, IgM-, and IgA-based antibody detection using enzyme-linked immunosorbent assay (ELISA). MATERIALS AND METHODS: We developed an indirect ELISA assay to quantitatively measure the amount of COVID-19 IgG, IgM, and IgA antibodies present in patient serum, dried blood, and plasma. RESULTS: The population cutoff values for positivity were determined by receiver operating characteristic curves to be 1.23 U/mL, 23.09 U/mL, and 6.36 U/mL for IgG, IgM, and IgA, respectively. After albumin subtraction, the specificity remained >98% and the sensitivity was 95.72%, 83.47%, and 82.60%, respectively, for IgG, IgM, and IgA antibodies to the combined spike subunit 1 receptor binding domain and N proteins in serum. Plasma and dried blood spot specimens were also validated on this assay. CONCLUSION: This assay may be used for determining the seroprevalence of SARS-CoV-2 in a population exposed to the virus or in vaccinated individuals.

Microsporidial Keratitis Related to Water Exposure: A Case Series.

The objective of this retrospective study was to present a series of cases involving the rare ocular disease of microsporidia keratitis treated at a private practice clinic and describe the details regarding specific water exposure, clinical course, voriconazole treatment, and increased prevalence of this infection in Augusta, Georgia, USA. Our analysis was based on the accumulated data from all patients (n=15) diagnosed with microsporidia keratitis at our private practice clinic; the clinical course of three cases is discussed in detail in this article. Specific environmental exposures were documented in 10 patients. All patients self-reported that they had no acquired immunodeficiency. In all cases, patients had complete resolution of active symptoms after receiving treatment with 1-2% topical voriconazole, with an approximate average primary treatment duration of 40.1 ± 17.1 days (median: 40 days, range: 14-70 days). None of the patients reported any clinically significant adverse effects from therapy. There have been increasing reports about this emerging infectious pathogen, particularly in Asia. However, there is limited data in the literature on the etiology, pathogenesis, and treatment of microsporidia-caused ophthalmic diseases. In this case series, we highlight the strong correlation of our patients' condition with specific types of water exposure in the USA as well as the complete resolution of active disease in all our patients as a result of monotherapy with topical voriconazole.

Comparison of Patient-Reported Functional Recovery From Different Types of Ophthalmic Surgery.

PURPOSE: To characterize and compare patient-reported recovery of function after cataract or glaucoma surgery using a novel visual analog scale. DESIGN: Prospective observational cohort study. METHODS: Daily for 2 weeks and weekly thereafter, patients recovering from trabeculectomy, tube shunt implantation, or cataract extraction (CE) completed a diary-style questionnaire including visual analog scales (VASs; scored 0-100) grading pain and global function. Clinical examination data and medical histories were collected. Generalized estimating equation models evaluated associations between VAS function scores and pain or visual acuity (VA) and compared scores between surgery types. RESULTS: Among 51 participants followed for 12 weeks, tube shunt placement reduced postoperative day 1 (POD1) function by 47 of 100 points vs CE (P = .006), while trabeculectomy did not reduce POD1 function vs CE (P = .33). After CE, trabeculectomy, and tube shunt placement, average VAS function scores increased 13.94 per week for 2 weeks (P < .001), 4.18 per week for 4 weeks (P = .02), and 7.76 per week for 7 weeks (P < .001), respectively. After those timepoints, there was no further significant change. Beyond 2 weeks, pain levels plateaued, and VA returned to baseline across surgery types; function was inversely related to pain or VA only for the first 2 or 4 weeks, respectively. CONCLUSIONS: Patients recovering from cataract and glaucoma surgery report reduced function in the postoperative period. Tube shunt implantation causes greater morbidity than trabeculectomy, and both are associated with slower improvement than CE. Early postoperative function is associated with VA and pain, but neither fully explains reported impairment. A VAS for function may efficiently capture postoperative recovery.

The Human TET2 Gene Contains Three Distinct Promoter Regions With Differing Tissue and Developmental Specificities.

Tet methylcytosine dioxygenase 2 (TET2) is a tumor suppressor gene that is inactivated in a wide range of hematological cancers. TET2 enzymatic activity converts 5-methylcytosine (5-mC) into 5-hydroxymethylcytosine (5-hmC), an essential step in DNA demethylation. Human TET2 is highly expressed in pluripotent cells and down-regulated in differentiated cells: however, transcriptional regulation of the human TET2 gene has not been investigated in detail. Here we define three promoters within a 2.5 kb region located ∼ 87 kb upstream of the first TET2 coding exon. The three promoters, designated as Pro1, Pro2, and Pro3, generate three alternative first exons, and their presence in TET2 mRNAs varies with cell type and developmental stage. In general, all three TET2 transcripts are more highly expressed in human tissues rich in hematopoietic stem cells, such as spleen and bone marrow, compared to other tissues, such as brain and kidney. Transcripts from Pro2 are expressed by a broad range of tissues and at a significantly higher level than Pro1 or Pro3 transcripts. Pro3 transcripts were highly expressed by embryoid bodies generated from the H9 ES cell line, and the major Pro3 transcript is an alternatively spliced mRNA isoform that produces a truncated TET2 protein lacking the catalytic domain. Our study demonstrates distinct tissue-specific mechanisms of TET2 transcriptional regulation during early pluripotent states and in differentiated cell types.

MATPLM1, A MATLAB script for scoring of periodic limb movements: preliminary validation with visual scoring.

BACKGROUND AND PURPOSE: A Matrix Laboratory (MATLAB) script (MATPLM1) was developed to rigorously apply World Associations of Sleep Medicine (WASM) scoring criteria for periodic limb movements in sleep (PLMS) from bilateral electromyographic (EMG) leg recordings. This study compares MATPLM1 with both standard technician and expert detailed visual PLMS scoring. METHODS AND SUBJECTS: Validation was based on a 'macro' level by agreement for PLMS/h during a night recording and on a 'micro' level by agreement for the detection of each PLMS from a stratified random sample for each subject. Data available for these analyses were from 15 restless leg syndrome (RLS) (age: 61.5 ± 8.5, 60% female) and nine control subjects (age: 61.4 ± 7.1, 67% female) participating in another study. RESULTS: In the 'micro' analysis, MATPLM1 and the visual detection of PLMS events agreed 87.7% for technician scoring and 94.4% for expert scoring. The technician and MATPLM1 scoring disagreements were checked for 36 randomly selected events, 97% involved clear technician-scoring error. In the 'macro' analysis, MATPLM1 rates of PMLS/h correlated highly with visual scoring by the technician (r(2) = 0.97) and the expert scorer (r(2) = 0.99), but the technician scoring was consistently less than MATPLM1: median (quartiles) difference: 10 (5, 23). There was little disagreement with expert scorer [median (quartile) difference: -0.3 (-2.4, 0.3)]. CONCLUSIONS: The MATPLM1 produces reliable scoring of PLMS that matches expert scoring. The standard visual scoring without careful measuring of events tends to significantly underscore PLMS. These preliminary results support the use of MATPLM1 as a preferred method of scoring PLMS for EMG recordings that are of a good quality and without significant sleep-disordered breathing events.