Tubular Injury Biomarkers to Predict Chronic Kidney Disease and Hypertension at 3-Months Post-Cisplatin in Children.

BACKGROUND: Urine kidney injury biomarkers measured during cisplatin therapy may identify patients at risk for adverse subsequent kidney outcomes. We examined relationships between tubular injury biomarkers collected early (early visit [EV]: first or second cisplatin cycle) and late (late visit [LV]: last or second-last cisplatin cycle) during cisplatin therapy, with 3-month post-cisplatin chronic kidney disease (CKD) and hypertension. METHODS: We analyzed data from the Applying Biomarkers to Minimize Long-Term Effects of Childhood/Adolescent Cancer Treatment Nephrotoxicity Study: twelve-center prospective cohort study of 159 children receiving cisplatin. We measured urine neutrophil gelatinase-associated lipocalin (NGAL)/creatinine, kidney injury molecule-1 (KIM-1)/creatinine, tissue inhibitor of metalloproteinase-2 (TIMP-2), and insulin-like growth factor-binding protein 7 (IGFBP-7) (TIMP-2 and IGFBP-7 expressed as their product, ng/ml^2/1000) at an EV and LV during cisplatin therapy with pre-infusion, post-infusion, and hospital discharge sampling. Area under the curve (AUC) was calculated for biomarkers to detect 3-month post-cisplatin CKD (KDIGO guidelines: low estimated glomerular filtration rate (eGFR) or elevated uACR for age) and hypertension (three blood pressures; per American Academy of Pediatrics guidelines). RESULTS: At median follow-up of 90 days, 52/118 (44%) and 17/125 (14%) developed CKD and hypertension, respectively. Biomarker prediction for 3-month CKD was low to modest; NGAL combined with KIM-1 at EV discharge yielded the highest AUC (0.67, 95% CI 0.57-0.77). Biomarker prediction of 3-month hypertension was stronger, but modest; the highest AUC was from combining EV pre-infusion NGAL and TIMP-2*IGFBP-7 (0.71, 95% CI 0.62-0.80). When EV pre-infusion NGAL and TIMP-2*IGFBP-7 were added to the 3-month hypertension clinical predictive model, AUCs increased from 0.81 (0.72-0.91) to 0.89 (0.83-0.95) (p<0.05). CONCLUSIONS: Tubular injury biomarkers we studied were individually not strong predictors of 3-month post-cisplatin kidney outcomes. Adding biomarkers to existing clinical prediction models may help predict post-therapy hypertension and identify higher kidney-risk patients.

Google Gemini and Bard artificial intelligence chatbot performance in ophthalmology knowledge assessment.

PURPOSE: With the popularization of ChatGPT (Open AI, San Francisco, California, United States) in recent months, understanding the potential of artificial intelligence (AI) chatbots in a medical context is important. Our study aims to evaluate Google Gemini and Bard's (Google, Mountain View, California, United States) knowledge in ophthalmology. METHODS: In this study, we evaluated Google Gemini and Bard's performance on EyeQuiz, a platform containing ophthalmology board certification examination practice questions, when used from the United States (US). Accuracy, response length, response time, and provision of explanations were evaluated. Subspecialty-specific performance was noted. A secondary analysis was conducted using Bard from Vietnam, and Gemini from Vietnam, Brazil, and the Netherlands. RESULTS: Overall, Google Gemini and Bard both had accuracies of 71% across 150 text-based multiple-choice questions. The secondary analysis revealed an accuracy of 67% using Bard from Vietnam, with 32 questions (21%) answered differently than when using Bard from the US. Moreover, the Vietnam version of Gemini achieved an accuracy of 74%, with 23 (15%) answered differently than the US version of Gemini. While the Brazil (68%) and Netherlands (65%) versions of Gemini performed slightly worse than the US version, differences in performance across the various country-specific versions of Bard and Gemini were not statistically significant. CONCLUSION: Google Gemini and Bard had an acceptable performance in responding to ophthalmology board examination practice questions. Subtle variability was noted in the performance of the chatbots across different countries. The chatbots also tended to provide a confident explanation even when providing an incorrect answer.

Conservation gaps and priorities of range-restricted birds in the Northern Andes.

The ongoing destruction of habitats in the tropics accelerates the current rate of species extinction. Range-restricted species are exceptionally vulnerable, yet we have insufficient knowledge about their protection. Species' current distributions, range sizes, and protection gaps are crucial to determining conservation priorities. Here, we identified priority range-restricted bird species and their conservation hotspots in the Northern Andes. We employed maps of the Area of Habitat (AOH), that better reflect their current distributions than existing maps. AOH provides unprecedented resolution and maps a species in the detail essential for practical conservation actions. We estimated protection within each species' AOH and for the cumulative distribution of all 335 forest-dependent range-restricted birds across the Northern Andes. For the latter, we also calculated protection across the elevational gradient. We estimated how much additional protection community lands (Indigenous and Afro-Latin American lands) would contribute if they were conservation-focused. AOHs ranged from 8 to 141,000 km2. We identified four conservation priorities based on cumulative species richness: the number of AOHs stacked per unit area. These priorities are high-resolution mapped representations of Endemic Bird Areas for the Tropical Andes that we consider critically important. Protected areas cover only 31% of the cumulative AOH, but community lands could add 19% more protection. Sixty-two per cent of the 335 species have ranges smaller than their published estimates, yet IUCN designates only 23% of these as Threatened. We identified 50 species as top conservation priorities. Most of these concentrate in areas of low protection near community lands and at middle elevations where, on average, only 34% of the land is protected. We highlight the importance of collaborative efforts among stakeholders: governments should support private and community-based conservation practices to protect the region with the most range-restricted birds worldwide.

Dose-response of localized renal cell carcinoma after stereotactic body radiation therapy: A meta-analysis.

BACKGROUND: Stereotactic ablative radiation therapy (SBRT) is an emerging treatment option for primary renal cell carcinoma (RCC), particularly in patients who are unsuitable for surgery. The aim of this review is to assess the effect of increasing the biologically equivalent dose (BED) via various radiation fractionation regimens on clinical outcomes. METHODS: A literature search was conducted in PubMed (Medline), EMBASE, and the Cochrane Library for studies published up to October 2023. Studies reporting on patients with localized RCC receiving SBRT were included to determine its effectiveness on local control, progression-free survival, and overall survival. A random effects model was used to meta-regress clinical outcomes relative to the BED for each study and heterogeneity was assessed by I2. RESULTS: A total of 724 patients with RCC from 22 studies were included, with a mean age of 72.7 years (range: 44.0-81.0). Local control was excellent with an estimate of 99 % (95 %CI: 97-100 %, I2 = 19 %), 98 % (95 %CI: 96-99 %, I2 = 8 %), and 94 % (95 %CI: 90-97 %, I2 = 11 %) at one year, two years, and five years respectively. No definitive association between increasing BED and local control, progression-free survival and overall survival was observed. No publication bias was observed. CONCLUSIONS: A significant dose response relationship between oncological outcomes and was not identified, and excellent local control outcomes were observed at the full range of doses. Until new evidence points otherwise, we support current recommendations against routine dose escalation beyond 25-26 Gy in one fraction or 42-48 Gy in three fractions, and to consider de-escalation or compromising target coverage if required to achieve safe organ at risk doses.

Accuracy of an Artificial Intelligence Chatbot's Interpretation of Clinical Ophthalmic Images.

Importance: Ophthalmology is reliant on effective interpretation of multimodal imaging to ensure diagnostic accuracy. The new ability of ChatGPT-4 (OpenAI) to interpret ophthalmic images has not yet been explored. Objective: To evaluate the performance of the novel release of an artificial intelligence chatbot that is capable of processing imaging data. Design, Setting, and Participants: This cross-sectional study used a publicly available dataset of ophthalmic cases from OCTCases, a medical education platform based out of the Department of Ophthalmology and Vision Sciences at the University of Toronto, with accompanying clinical multimodal imaging and multiple-choice questions. Across 137 available cases, 136 contained multiple-choice questions (99%). Exposures: The chatbot answered questions requiring multimodal input from October 16 to October 23, 2023. Main Outcomes and Measures: The primary outcome was the accuracy of the chatbot in answering multiple-choice questions pertaining to image recognition in ophthalmic cases, measured as the proportion of correct responses. χ2 Tests were conducted to compare the proportion of correct responses across different ophthalmic subspecialties. Results: A total of 429 multiple-choice questions from 136 ophthalmic cases and 448 images were included in the analysis. The chatbot answered 299 of multiple-choice questions correctly across all cases (70%). The chatbot's performance was better on retina questions than neuro-ophthalmology questions (77% vs 58%; difference = 18%; 95% CI, 7.5%-29.4%; χ21 = 11.4; P < .001). The chatbot achieved a better performance on nonimage-based questions compared with image-based questions (82% vs 65%; difference = 17%; 95% CI, 7.8%-25.1%; χ21 = 12.2; P < .001).The chatbot performed best on questions in the retina category (77% correct) and poorest in the neuro-ophthalmology category (58% correct). The chatbot demonstrated intermediate performance on questions from the ocular oncology (72% correct), pediatric ophthalmology (68% correct), uveitis (67% correct), and glaucoma (61% correct) categories. Conclusions and Relevance: In this study, the recent version of the chatbot accurately responded to approximately two-thirds of multiple-choice questions pertaining to ophthalmic cases based on imaging interpretation. The multimodal chatbot performed better on questions that did not rely on the interpretation of imaging modalities. As the use of multimodal chatbots becomes increasingly widespread, it is imperative to stress their appropriate integration within medical contexts.

Diagnostic methods for primary vitreoretinal lymphoma: A systematic review.

Primary vitreoretinal lymphoma is a potentially aggressive intraocular malignancy with poor systemic prognosis and sometimes significant diagnostic delays as it may masquerade as chronic uveitis. Despite the variety of diagnostic techniques, it is unclear which modality is most accurate in the diagnosis of PVRL. A systematic literature search was conducted on Ovid MEDLINE, EMBASE and the Cochrane Controlled Register of Trials for studies published between January, 2000, and June, 2023. Randomized controlled trials (RCTs) reporting on the following diagnostic tools used to diagnose patients with PVRL were included: cytology, flow cytometry, MYD88 L265P mutation, CD79B mutation, interleukin 10/interleukin-6 (IL-10/IL-6) ratio, polymerase chain reaction (PCR) for monoclonal immunoglobulin heavy chain (IgH) and immunoglobulin kappa light chain (IgK) rearrangements, and imaging findings. The aggregated sensitivity of each diagnostic modality was reported and compared using the chi-squared (χ2) test. A total of 662 eyes from 29 retrospective studies reporting on patients diagnosed with PVRL were included. An IL-10/IL-6 ratio greater than 1 had the highest sensitivity (89.39%, n = 278/311 eyes, n = 16 studies) for PVRL, where the sensitivity was not significantly different when only vitreous samples were drawn (88.89%, n = 232/261 eyes, n = 13 studies) compared to aqueous samples (83.33%, n = 20/24, n = 2) (p = 0.42). Flow cytometry of vitreous samples gave a positive result in 66/75 eyes (88.00%, n = 6 studies) with PVRL, and monoclonal IgH rearrangements on PCR gave a positive result in 354/416 eyes (85.10%, n = 20 studies) with PVRL. MYD88 L265P and CD79B mutation analysis performed poorly, yielding a positive result in 63/90 eyes (70.00%, n = 8 studies) with PVRL, and 20/57 eyes (35.09%, n = 4 studies) with PVRL, respectively. Overall, our systematic review found that an IL-10/IL-6 ratio greater or equal to one may provide the highest sensitivity in identifying patients with PVRL. Future studies are needed to employ multiple diagnostic tools to aid in the detection of PVRL and to further establish nuanced guidelines when determining the optimal diagnostic tool to use in diverse patient populations.

Cleft lip and/or palate mortality trends in the USA: a retrospective population-based study.

BACKGROUND: Cleft lip and/or palate (CL/P) is one of the most common congenital anomalies worldwide. Although CL/P management may require a series of interventions, mortality resulting from CL/P alone is rare. This study aims to examine recent trends of CL/P mortality rates in the USA. METHODS: A retrospective population-based study was conducted using official US birth and death certificate data from the Centers for Disease Control and Prevention from 2000 to 2019. Annual mortality rates per 1000 births with CL/P were calculated across sex and racial groups. Multivariable logistic regression models estimated the effects of sex and race on the risk of mortality with CL/P, and linear regression models were used to examine temporal changes in mortality rate across sex and race. RESULTS: From 2000 to 2019, 1119 deaths occurred in patients with documented CL/P, for an overall incidence of 20.3 deaths per 1000 births with CL/P (95% CI 18.9 to 22.8). Of these, Patau syndrome was the listed cause of death in 167 cases (14.9%). Black individuals (OR 1.93, 95% CI 1.85 to 2.01), Hispanic (1.54, 1.49 to 1.58) and American Indian individuals (1.28, 1.20 to 1.35) were at a greater risk of CL/P mortality compared with white individuals. Additionally, females were also at a greater risk (1.35, 1.21 to 1.49). A significant upward trend in CL/P mortality was observed in Hispanic (r2=0.70, p<0.01) and American Indian individuals (r2=0.81, p<0.01) from 2000 to 2019. CONCLUSIONS: Cleft birth and mortality surveillance is essential in healthcare and prevention planning. Future studies are required to understand the differences in CL/P mortality rates across various sociodemographic groups.

PARS PLANA VITRECTOMY WITH OR WITHOUT INTERNAL LIMITING MEMBRANE PEEL FOR MACULAR HOLE: A Systematic Review and Meta-Analysis.

PURPOSE: To compare the efficacy and safety of pars plana vitrectomy with and without internal limiting membrane (ILM) peeling for macular hole (MH). METHODS: A systematic literature search on Ovid MEDLINE, Embase, Cochrane Library, and Google Scholar was performed from January 2000 to 2023. The primary outcome was the final best-corrected visual acuity (BCVA). Secondary outcomes included MH closure rates and the need for repeat surgery. The authors performed a random-effects meta-analysis on Review Manager 5.4. RESULTS: Fourteen studies on 880 eyes were included. Pars plana vitrectomy with and without ILM peel achieved a similar final BCVA ( P = 0.66). However, pars plana vitrectomy without ILM peeling achieved a significantly better final BCVA in eyes with closed MHs (WMD = 0.05 logMAR, 95% CI, 0.01-0.10, P = 0.02). Pars plana vitrectomy with ILM peeling achieved a significantly higher primary MH closure rate (RR = 1.21, 95% CI, 1.04-1.42, P = 0.02) and lower incidence of MH reoperation (RR = 0.19, 95% CI, 0.11-0.33, P < 0.001). The final MH closure rate ( P = 0.12) and incidence of MH recurrence ( P = 0.25) were similar between groups. CONCLUSION: Pars plana vitrectomy with and without ILM peel achieved a similar final BCVA. However, pars plana vitrectomy without ILM peeling achieved a better final BCVA in eyes with closed MHs. ILM peeling achieved a greater primary MH closure rate and reduced need for reoperation.

Generative modeling of biological shapes and images using a probabilistic α-shape sampler.

Understanding morphological variation is an important task in many areas of computational biology. Recent studies have focused on developing computational tools for the task of sub-image selection which aims at identifying structural features that best describe the variation between classes of shapes. A major part in assessing the utility of these approaches is to demonstrate their performance on both simulated and real datasets. However, when creating a model for shape statistics, real data can be difficult to access and the sample sizes for these data are often small due to them being expensive to collect. Meanwhile, the current landscape of generative models for shapes has been mostly limited to approaches that use black-box inference-making it difficult to systematically assess the power and calibration of sub-image models. In this paper, we introduce the α-shape sampler: a probabilistic framework for generating realistic 2D and 3D shapes based on probability distributions which can be learned from real data. We demonstrate our framework using proof-of-concept examples and in two real applications in biology where we generate (i) 2D images of healthy and septic neutrophils and (ii) 3D computed tomography (CT) scans of primate mandibular molars. The α-shape sampler R package is open-source and can be downloaded at https://github.com/lcrawlab/ashapesampler.

The ability of artificial intelligence chatbots ChatGPT and Google Bard to accurately convey pre-operative information for patients undergoing ophthalmological surgeries.

INTRODUCTION: To determine whether the two popular artificial intelligence (AI) chatbots, ChatGPT and Bard, provide high-quality information concerning procedure description, risks, benefits, and alternatives of various ophthalmological surgeries. METHODS: ChatGPT and Bard were prompted with questions pertaining to the description, potential risks, benefits, alternatives, and implications of not proceeding with various surgeries in different subspecialties of ophthalmology. Six common ophthalmic procedures were included in our analysis. Two comprehensive ophthalmologists and one sub-specialist graded each response independently using a five-point Likert scale. RESULTS: Likert grading for accuracy was significantly higher for ChatGPT in comparison to Bard (4.5±0.6 vs 3.8±0.8, p<0.0001). Generally, ChatGPT performed better than Bard even when questions were stratified by type of ophthalmological surgery. There was no significant difference between ChatGPT and Bard for response length (2104.7±271.4 characters vs 2441.0±633.9 characters, p=0.12). ChatGPT responded significantly slower than Bard (46.0±3.0 seconds vs 6.6±1.2 seconds, p<0.0001). CONCLUSIONS: Both ChatGPT and Bard may offer accessible and high-quality information relevant to the informed consent process for various ophthalmic procedures. Nonetheless, both AI chatbots overlooked probability of adverse events, hence limiting their potential and introducing patients to information that may be difficult to interpret.

Protecting and connecting landscapes stabilizes populations of the Endangered savannah elephant.

The influence of protected areas on the growth of African savannah elephant populations is inadequately known. Across southern Africa, elephant numbers grew at 0.16% annually for the past quarter century. Locally, much depends on metapopulation dynamics-the size and connections of individual populations. Population numbers in large, connected, and strictly protected areas typically increased, were less variable from year to year, and suffered less from poaching. Conversely, populations in buffer areas that are less protected but still connected have more variation in growth from year to year. Buffer areas also differed more in their growth rates, likely due to more threats and dispersal opportunities in the face of such dangers. Isolated populations showed consistently high growth due to a lack of emigration. This suggests that "fortress" conservation generally maintains high growth, while anthropogenic-driven source-sink dynamics within connected conservation clusters drive stability in core areas and variability in buffers.

Artificial Intelligence Chatbots' Understanding of the Risks and Benefits of Computed Tomography and Magnetic Resonance Imaging Scenarios.

PURPOSE: Patients may seek online information to better understand medical imaging procedures. The purpose of this study was to assess the accuracy of information provided by 2 popular artificial intelligence (AI) chatbots pertaining to common imaging scenarios' risks, benefits, and alternatives. METHODS: Fourteen imaging-related scenarios pertaining to computed tomography (CT) or magnetic resonance imaging (MRI) were used. Factors including the use of intravenous contrast, the presence of renal disease, and whether the patient was pregnant were included in the analysis. For each scenario, 3 prompts for outlining the (1) risks, (2) benefits, and (3) alternative imaging choices or potential implications of not using contrast were inputted into ChatGPT and Bard. A grading rubric and a 5-point Likert scale was used by 2 independent reviewers to grade responses. Prompt variability and chatbot context dependency were also assessed. RESULTS: ChatGPT's performance was superior to Bard's in accurately responding to prompts per Likert grading (4.36 ± 0.63 vs 3.25 ± 1.03 seconds, P < .0001). There was substantial agreement between independent reviewer grading for ChatGPT (κ = 0.621) and Bard (κ = 0.684). Response text length was not statistically different between ChatGPT and Bard (2087 ± 256 characters vs 2162 ± 369 characters, P = .24). Response time was longer for ChatGPT (34 ± 2 vs 8 ± 1 seconds, P < .0001). CONCLUSIONS: ChatGPT performed superior to Bard at outlining risks, benefits, and alternatives to common imaging scenarios. Generally, context dependency and prompt variability did not change chatbot response content. Due to the lack of detailed scientific reasoning and inability to provide patient-specific information, both AI chatbots have limitations as a patient information resource.

Fc gamma receptors promote antibody-induced LILRB4 internalization and immune regulation of monocytic AML.

The immune checkpoint leukocyte immunoglobulin-like receptor B4 (LILRB4) is found specifically on the cell surface of acute monocytic leukemia (monocytic AML), an aggressive and common subtype of AML. We have developed a humanized monoclonal IgG1 LILRB4-blocking antibody (h128-3), which improved immune regulation but reduced cell surface expression of LILRB4 in monocytic AML models by 40-60%. Interestingly, most of this effect was neutralized by mutation of the Fc region of the antibody (h128-3/N297A), which prevents interaction with Fc gamma receptors (FcγRs). This suggested that there is FcγR-dependent antigenic modulation underlying h128-3's effects, a mechanism known to alter the function of antibodies targeting B-cell malignancies. We disrupted the Fc-FcγR interaction pharmacologically and with stable CRISPR-Cas9-mediated genetic knockout of FcγRs in monocytic AML cell lines to investigate the role of FcγR-dependent antigenic modulation in the regulation of LILRB4 by h128-3. When FcγRI is inhibited or removed from the surface of monocytic AML cells, h128-3 cannot optimally perform its blocking function, resulting in activation of the LILRB4 inhibitory receptor and leading to a 15-25% decrease in T-cell-mediated cytotoxicity in vitro. In the absence of FcγRI, scaffolding by FcγRIIa allows h128-3 to maintain LILRB4-blocking function. Here we define a FcγR-dependent antigenic modulation mechanism underlying the function of an immunoreceptor blocking antibody for the first time in myeloid malignancy. This research will facilitate the development of safe, precision-targeted antibody therapeutics in myeloid malignancies with greater potency and efficacy.

Racial disparities in national maternal mortality trends in the United States from 2000 to 2019: a population-based study on 80 million live births.

PURPOSE: In the United States (US), deaths during pregnancy and childbirth have increased over the past 2 decades compared to other high-income countries, and there have been reports that racial disparities in maternal mortality have widened. The study objective was to examine recent trends in maternal mortality in the US by race. METHODS: Our population-based cross-sectional study used data from the Centers for Disease Control and Prevention's 2000-2019 "Birth Data" and "Mortality Multiple Cause" data files from the US to calculate maternal mortality during pregnancy, childbirth, and puerperium across race. Logistic regression models estimated the effects of race on the risk of maternal mortality and examined temporal changes in risk across race. RESULTS: A total of 21,241 women died during pregnancy and childbirth, with 65.5% caused by obstetrical complications and 34.5% by non-obstetrical causes. Black women, compared with White women, had greater risk of maternal mortality (OR 2.13, 95% CI 2.06-2.20), as did American Indian women (2.02, 1.83-2.24). Overall maternal mortality risk increased during the 20-year study period, with an annual increase of 2.4 and 4.7/100,000 among Black and American Indian women, respectively. CONCLUSIONS: Between 2000 and 2019, maternal mortality in the US increased, overall and especially in American Indian and Black women. Targeted public health interventions to improve maternal health outcomes should become a priority.

ChatGPT-4: An assessment of an upgraded artificial intelligence chatbot in the United States Medical Licensing Examination.

PURPOSE: ChatGPT-4 is an upgraded version of an artificial intelligence chatbot. The performance of ChatGPT-4 on the United States Medical Licensing Examination (USMLE) has not been independently characterized. We aimed to assess the performance of ChatGPT-4 at responding to USMLE Step 1, Step 2CK, and Step 3 practice questions. METHOD: Practice multiple-choice questions for the USMLE Step 1, Step 2CK, and Step 3 were compiled. Of 376 available questions, 319 (85%) were analyzed by ChatGPT-4 on March 21st, 2023. Our primary outcome was the performance of ChatGPT-4 for the practice USMLE Step 1, Step 2CK, and Step 3 examinations, measured as the proportion of multiple-choice questions answered correctly. Our secondary outcomes were the mean length of questions and responses provided by ChatGPT-4. RESULTS: ChatGPT-4 responded to 319 text-based multiple-choice questions from USMLE practice test material. ChatGPT-4 answered 82 of 93 (88%) questions correctly on USMLE Step 1, 91 of 106 (86%) on Step 2CK, and 108 of 120 (90%) on Step 3. ChatGPT-4 provided explanations for all questions. ChatGPT-4 spent 30.8 ± 11.8 s on average responding to practice questions for USMLE Step 1, 23.0 ± 9.4 s per question for Step 2CK, and 23.1 ± 8.3 s per question for Step 3. The mean length of practice USMLE multiple-choice questions that were answered correctly and incorrectly by ChatGPT-4 was similar (difference = 17.48 characters, SE = 59.75, 95%CI = [-100.09,135.04], t = 0.29, p = 0.77). The mean length of ChatGPT-4's correct responses to practice questions was significantly shorter than the mean length of incorrect responses (difference = 79.58 characters, SE = 35.42, 95%CI = [9.89,149.28], t = 2.25, p = 0.03). CONCLUSIONS: ChatGPT-4 answered a remarkably high proportion of practice questions correctly for USMLE examinations. ChatGPT-4 performed substantially better at USMLE practice questions than previous models of the same AI chatbot.