Correction: How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment.

[This corrects the article DOI: 10.2196/45312.].

How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment.

BACKGROUND: Chat Generative Pre-trained Transformer (ChatGPT) is a 175-billion-parameter natural language processing model that can generate conversation-style responses to user input. OBJECTIVE: This study aimed to evaluate the performance of ChatGPT on questions within the scope of the United States Medical Licensing Examination (USMLE) Step 1 and Step 2 exams, as well as to analyze responses for user interpretability. METHODS: We used 2 sets of multiple-choice questions to evaluate ChatGPT's performance, each with questions pertaining to Step 1 and Step 2. The first set was derived from AMBOSS, a commonly used question bank for medical students, which also provides statistics on question difficulty and the performance on an exam relative to the user base. The second set was the National Board of Medical Examiners (NBME) free 120 questions. ChatGPT's performance was compared to 2 other large language models, GPT-3 and InstructGPT. The text output of each ChatGPT response was evaluated across 3 qualitative metrics: logical justification of the answer selected, presence of information internal to the question, and presence of information external to the question. RESULTS: Of the 4 data sets, AMBOSS-Step1, AMBOSS-Step2, NBME-Free-Step1, and NBME-Free-Step2, ChatGPT achieved accuracies of 44% (44/100), 42% (42/100), 64.4% (56/87), and 57.8% (59/102), respectively. ChatGPT outperformed InstructGPT by 8.15% on average across all data sets, and GPT-3 performed similarly to random chance. The model demonstrated a significant decrease in performance as question difficulty increased (P=.01) within the AMBOSS-Step1 data set. We found that logical justification for ChatGPT's answer selection was present in 100% of outputs of the NBME data sets. Internal information to the question was present in 96.8% (183/189) of all questions. The presence of information external to the question was 44.5% and 27% lower for incorrect answers relative to correct answers on the NBME-Free-Step1 (P<.001) and NBME-Free-Step2 (P=.001) data sets, respectively. CONCLUSIONS: ChatGPT marks a significant improvement in natural language processing models on the tasks of medical question answering. By performing at a greater than 60% threshold on the NBME-Free-Step-1 data set, we show that the model achieves the equivalent of a passing score for a third-year medical student. Additionally, we highlight ChatGPT's capacity to provide logic and informational context across the majority of answers. These facts taken together make a compelling case for the potential applications of ChatGPT as an interactive medical education tool to support learning.

Discovery and Characterization of Potent, Efficacious, Orally Available Antimalarial Plasmepsin X Inhibitors and Preclinical Safety Assessment of UCB7362.

Plasmepsin X (PMX) is an essential aspartyl protease controlling malaria parasite egress and invasion of erythrocytes, development of functional liver merozoites (prophylactic activity), and blocking transmission to mosquitoes, making it a potential multistage drug target. We report the optimization of an aspartyl protease binding scaffold and the discovery of potent, orally active PMX inhibitors with in vivo antimalarial efficacy. Incorporation of safety evaluation early in the characterization of PMX inhibitors precluded compounds with a long human half-life (t1/2) to be developed. Optimization focused on improving the off-target safety profile led to the identification of UCB7362 that had an improved in vitro and in vivo safety profile but a shorter predicted human t1/2. UCB7362 is estimated to achieve 9 log 10 unit reduction in asexual blood-stage parasites with once-daily dosing of 50 mg for 7 days. This work demonstrates the potential to deliver PMX inhibitors with in vivo efficacy to treat malaria.

Quantitative assessment of bony orbital volume symmetry: CT analysis in the uninjured caucasian population.

There is a surprising lack of evidence documenting the volumetric symmetry of the bony orbit. This paper establishes reference values for orbital volume (OV) and symmetry in the 25 - 40 year old caucasian population. Secondarily, this paper sets a landmark for the tolerances in OV that can be expected when reconstructing the bony defects which may occur from trauma. A standardised method of quantitative OV measurement was developed using CT sinus examinations acquired for indications unrelated to orbital trauma. Sex, ethnicity, age, right and left OV were recorded. Data for 100 patients was obtained (50 male, 50 female). Mean left OV was 23.1cm3 and mean right OV was 23.3cm3. Left and right OV were strongly positively correlated (correlation coefficient: 0.96). Mean female OV was 21.6cm3 and mean male OV was 24.8cm3. On average, male OV is 3.2cm3 larger than female OV. The mean difference between left and right OV was 0.5cm3 in females and 0.6cm3 in males. The intra-class coefficient score between the two assessors was 0.973 (excellent). There is strong positive correlation between left and right OV in this study population. Previous work suggests that orbital volume loss less than 1cm3 would not lead to significant clinical symptoms of orbital fracture. When orbital reconstruction is undertaken, this study suggests that a volume symmetry difference of <0.5cm3 in females and <0.6cm3 in males would be consistent with the variation seen in the study population of uninjured caucasian 25-40 year olds and is therefore a reasonable goal of surgical management.

COVID-19 and Pediatric Gastroenterology.

Pediatric gastroenterologists took on a variety of challenges during the coronavirus disease 2019 pandemic, including learning about a new disease and how to recognize and manage it, prevent its spread among their patients and health professions colleagues, and make decisions about managing patients with chronic gastrointestinal and liver problems in light of the threat. They adapted their practice to accommodate drastically decreased numbers of in-person visits, adopting telehealth technologies, and instituting new protocols to perform endoscopies safely. The workforce pipeline was also affected by the impact of the pandemic on trainee education, clinical experience, research, and job searches.

Membrane Interactions of α-Synuclein Revealed by Multiscale Molecular Dynamics Simulations, Markov State Models, and NMR.

α-Synuclein (αS) is a presynaptic protein that binds to cell membranes and is linked to Parkinson's disease (PD). Binding of αS to membranes is a likely first step in the molecular pathophysiology of PD. The αS molecule can adopt multiple conformations, being largely disordered in water, adopting a ß-sheet conformation when present in amyloid fibrils, and forming a dynamic multiplicity of α-helical conformations when bound to lipid bilayers and related membrane-mimetic surfaces. Multiscale molecular dynamics simulations in conjunction with nuclear magnetic resonance (NMR) and cross-linking mass spectrometry (XLMS) measurements are used to explore the interactions of αS with an anionic lipid bilayer. The simulations and NMR measurements together reveal a break in the helical structure of the central non-amyloid-ß component (NAC) region of αS in the vicinity of residues 65-70, which may facilitate subsequent oligomer formation. Coarse-grained simulations of αS starting from the structure of αS when bound to a detergent micelle reveal the overall pattern of protein contacts to anionic lipid bilayers, while subsequent all-atom simulations provide details of conformational changes upon membrane binding. In particular, simulations and NMR data for liposome-bound αS indicate incipient ß-strand formation in the NAC region, which is supported by intramolecular contacts seen via XLMS and simulations. Markov state models based on the all-atom simulations suggest a mechanism of conformational change of membrane-bound αS via a dynamic helix break in the region of residue 65 in the NAC region. The emergent dynamic model of membrane-interacting αS advances our understanding of the mechanism of PD, potentially aiding the design of novel therapeutic approaches.

A cue-to-action pilot project to increase screening mammography.

OBJECTIVES: Screening mammography is a preventive exam used to detect breast cancer in asymptomatic women. This cue-to-action pilot project sought to determine if outreach, education, and incentive would increase uptake of screening mammography among women aged 52 to 74 years who are members of a community-based health insurance plan. STUDY DESIGN: Cohort study design with retrospective and prospective components. METHODS: Women were eligible to participate in accordance with the CMS Quality Rating System technical specification for breast cancer screening. Eligible women with no documented screening for a mammogram from January 1, 2016, through November 7, 2017, were invited to participate in a campaign that included outreach about screening mammography as a no-cost covered benefit, education about screening mammography to detect asymptomatic disease, and a gift card to a local grocery merchant if the member obtained screening mammography by December 31, 2017. RESULTS: During December 2017, 20.8% (27/130) of eligible women obtained a screening mammogram compared with 7.8% (5/64) of eligible women during the nonintervention reference period of December 2016. Mammography screening increased by 170% during the study period in comparison with the reference period of a year earlier (prevalence ratio [PR], 2.7; 95% CI, 1.1-6.6; P = .02). CONCLUSIONS: A one-time, time-limited cue-to-action pilot project consisting of outreach, education, and incentive increased uptake of screening mammography by women enrolled in a community health insurance plan providing health insurance coverage as part of the Affordable Care Act. This increase is statistically significant in the intervention period compared with the reference period (PR, 2.7; 95% CI, 1.1-6.6; P = .02). Despite a small sample size, the magnitude of the effect for this pilot study is encouraging and warrants future studies in a larger population.

Hepatitis C Antibody Screening Among Baby Boomers by a Community-Based Health Insurance Company.

Infection with the hepatitis C virus (HCV) is the most commonly reported bloodborne infection in the United States. Individuals born between 1945-1965, the baby boomers, account for approximately 75% of all chronic HCV infections in the United States. The purpose of this study was to determine if a 6-week intervention, including outreach, education, and incentive, by a community-based health insurance company could improve uptake of HCV antibody screening among the 1945-1965 birth cohort. Individuals were eligible to participate in this campaign if they were born on or after January 1, 1945 and on or before December 31, 1965, had health insurance with Sendero Health Plans during the intervention period, and had no evidence of having received an HCV antibody test prior to the campaign start date. The 6-week campaign period was from November 14, 2018 through December 31, 2018. A gift card incentive was provided if HCV screening was completed on or before December 31, 2018. A total of 5287 individuals were eligible to participate in the campaign. Members who were baby boomers were 3.36 times more likely to receive HCV antibody screening during the intervention period in 2018 than during a similar period in 2017 (prevalence ratio = 3.36; P < 0.0001; 95% confidence interval: 2.71, 4.16). Health officials have established the identification, treatment, and elimination of HCV as a national policy objective. Using an outreach, education, and incentive approach, Sendero Health Plans improved uptake of HCV antibody screening among the high-risk baby boomer population.

A Randomized Controlled Trial to Increase Diabetic Retinopathy Screening by a Community-Based Health Insurance Plan in Central Texas - 2017.

PURPOSE: Diabetic retinopathy is a leading cause of blindness worldwide. In the United States, the prevalence of diabetic retinopathy is 26% - 33%. Providing preventive care to individuals with diabetes is important to prevent microvascular complications in the eye. This study reports on the results of a randomized controlled trial to determine how using specific cues to action combined with the provision of a free eye exam might positively influence the rate of diabetic retinopathy screening among individuals with diabetes. METHODS: Individuals were eligible to participate in this campaign if they had a diagnosis of type 2 diabetes or were prescribed a diabetes drug, were members of the health insurance plan during the intervention period and had no evidence of receiving a retinal eye exam prior to the campaign period. The six-week campaign period started on September 19, 2017 and ended on October 31, 2017. A total of 1,454 individuals with type 2 diabetes were randomly assigned to a control group or to one of three intervention groups. Each intervention group included the provision of a free eye exam. RESULTS: A total of 148 (10.1%) individuals obtained a retinal eye exam during the six-week campaign period with 38 persons (6.8%) in the control group, 38 (15.3%) in the incentive group, 41 (16.5%) in the education group, and 31 (12.4%) in the incentive and education group. Individual intervention comparisons with the referent group yielded statistical significance using the adjusted pairwise alpha of P = .008 for the incentive group (RR = 2.08; 95% CI, 1.36-3.19; P =.0006) and for the education group (RR = 2.23; 95% CI, 1.47-3.39; P =.0001), but not in the incentive plus education group (RR = 1.73; 95% CI, 1.10-2.73; P =.017). CONCLUSION: This study supports the use of targeted cues to action combined with the provision of a free eye exam to increase the rates of diabetic retinopathy screening among individuals with diabetes who have health insurance coverage under the Affordable Care Act in a Central Texas population.

Using Health Insurance Network Provider Data and Public Data Sets to Identify SARS-CoV-2 Vaccinators in the USA.

Objective: Mass vaccination planning is occurring at all levels of government in advance of regulatory approval and manufacture of a SARS-CoV-2 vaccine for distribution sometime in 2021. We outline a methodology in which both health insurance provider network data and publicly available data sources can be used to identify and plan for SARS-CoV-2 vaccinator capacity at the county level. Methods: Sendero Health Plans, Inc. provider network data, Texas State Board of Pharmacy data, US Census Bureau data, and H1N1 monovalent vaccine data were utilized to identify providers with demonstrated capacity to vaccinate the population in Travis County, Texas to achieve an estimated SARS-CoV-2 herd immunity target of 67%. Results: Within the Sendero network, 2,356 non-pharmacy providers were identified with 788 (33.4%) practicing in primary care and 1,569 (66.6%) practicing as specialists. Of the total, 686 (29.1%) provided at least one immunization between January 1, 2019 and September 30, 2020. There are 300 pharmacies with active licenses in Travis County with 161 (53.7%) classified as community pharmacies. We estimate that 1,707,098 doses of a 2-dose SARS-CoV-2 vaccine series will need to be administered within Travis County, Texas to achieve the estimated 67% herd immunity threshold to disrupt person-to-person transmission of the SARS-CoV-2 virus based on 2020 census data. Conclusion: A community-based health insurance plan can use data from its provider network and public data sources to support the CDC call to action to identify SARS-CoV-2 vaccinators in the community, including physicians, nurse practitioners, physician assistants, and pharmacies in order to provide macro level estimates of SARS-CoV-2 administration and throughput.

Emergency department encounters for opioid abuse, adverse events, poisoning, and dependence among members of a community-based health insurance plan-Central Texas, 2016-2018.

BACKGROUND: The United States appears to be in the midst of an opioid epidemic. National data indicate a rise in emergency department visits for opioid-related causes over the past decade. This data, while important in helping to explain the magnitude of the epidemic nationally offers only a glimpse of what can be expected to occur locally. The objective of this secondary data analysis was to describe the impact that opioid abuse, adverse events, poisoning, and dependence have on emergency department utilization for individuals who purchased health insurance under the Affordable Care Act in Central Texas from a community-based health maintenance organization. METHODS: Individuals who purchased health insurance from Sendero Health Plans in calendar years 2016, 2017, and 2018 were eligible for participation if they had both an emergency department encounter and an opioid-related ICD-10-CM diagnosis. Eligible individuals were assessed to determine if they were dispensed an opioid agonist or opioid antagonist prescription during the year of their emergency department encounter. Sendero medical claims data for calendar years 2016, 2017, and 2018 were used to calculate both the incidence and ratio of emergency department visits per 100,000-person Sendero member population. Sendero data were compared to available national data estimates. RESULTS: A total of 55 individuals had an emergency department encounter with a primary or secondary opioid-related diagnosis from January 1, 2016 through December 31, 2018. These 55 individuals had 69 unique emergency department encounters during this time period. The incidence of new claims per 100,000-member Sendero population was 67.1, 64.5, and 62.6 in 2016, 2017, and 2018 respectively. The ratio of unique emergency department encounters per 100,000-member Sendero population was 95.9, 82.6, and 66.5 in 2016, 2017, and 2018 respectively. CONCLUSION: Health insurance claims data from a community-based health plan can be used as a source of local information by policy makers and officials as they seek to address the impact of opioid abuse, adverse events, poisoning, and dependence in Central Texas as national data may not represent the local impact of this epidemic.

Dietary Glycemic Index and Load and the Risk of Type 2 Diabetes: Assessment of Causal Relations.

While dietary factors are important modifiable risk factors for type 2 diabetes (T2D), the causal role of carbohydrate quality in nutrition remains controversial. Dietary glycemic index (GI) and glycemic load (GL) have been examined in relation to the risk of T2D in multiple prospective cohort studies. Previous meta-analyses indicate significant relations but consideration of causality has been minimal. Here, the results of our recent meta-analyses of prospective cohort studies of 4 to 26-y follow-up are interpreted in the context of the nine Bradford-Hill criteria for causality, that is: (1) Strength of Association, (2) Consistency, (3) Specificity, (4) Temporality, (5) Biological Gradient, (6) Plausibility, (7) Experimental evidence, (8) Analogy, and (9) Coherence. These criteria necessitated referral to a body of literature wider than prospective cohort studies alone, especially in criteria 6 to 9. In this analysis, all nine of the Hill's criteria were met for GI and GL indicating that we can be confident of a role for GI and GL as causal factors contributing to incident T2D. In addition, neither dietary fiber nor cereal fiber nor wholegrain were found to be reliable or effective surrogate measures of GI or GL. Finally, our cost-benefit analysis suggests food and nutrition advice favors lower GI or GL and would produce significant potential cost savings in national healthcare budgets. The high confidence in causal associations for incident T2D is sufficient to consider inclusion of GI and GL in food and nutrient-based recommendations.

Disparities in Hospice Utilization for Older Cancer Patients Living in the Deep South.

CONTEXT: Hospice utilization is an end-of-life quality indicator. The Deep South has known disparities in palliative care that may affect hospice utilization. OBJECTIVES: The objective of this study was to evaluate the association among Deep South patient and hospital characteristics and hospice utilization. METHODS: This retrospective cohort study evaluated patient and hospital characteristics associated with hospice among Medicare cancer decedents aged ≥65 years in 12 southeastern cancer centers between 2012 and 2015. We examined patient-level characteristics (age, race, gender, cancer type, and received patient navigation) and hospital-level characteristics (board-certified palliative physician, inpatient palliative care beds, and hospice ownership). Outcomes included hospice (within 90 vs. three days of death). Relative risks (RRs) and 95% CIs evaluated the association between patient- and hospital-level characteristics and hospice outcomes using generalized log-linear models with Poisson distribution and robust variance estimates. RESULTS: Of 12,725 cancer decedents, 4142 (33%) did not utilize hospice. "No hospice" was associated with nonwhite (RR 1.24, 95% CI 1.17-1.32) and nonnavigated patients (RR 1.17, 95% CI 1.10-1.25), and those at a hospital with inpatient palliative care beds (RR 1.15, 95% CI 1.10-1.21). "Late hospice" (20%; n = 1458) was associated with being male (RR 1.31, 95% CI 1.19-1.44) and seen at a hospital without inpatient palliative care beds (RR 0.82, 95% CI 0.75-0.90). CONCLUSIONS: Hospice utilization differed by patient and hospital characteristics. Patients who were nonwhite, and nonnavigated, and hospitals with inpatient palliative care beds, were associated with no hospice. Research should focus on ways to improve hospice utilization in Deep South older cancer patients.

Lay navigation across the cancer continuum for older cancer survivors: Equally beneficial for Black and White survivors?

OBJECTIVES: The Patient Care Connect Program (PCCP), through lay navigators' distress assessments and assistance, was shown to lower healthcare utilization and costs in older cancer survivors. PCCP benefits and assistance needs for disadvantaged minorities (e.g., Black) vs. Whites are unknown. MATERIALS AND METHODS: We examined the PCCP impact with retrospective analyses of Medicare claims (2012-2015). Outcomes were quarterly costs and utilization (emergency room (ER), hospitalizations) for navigated and matched survivors. Repeated measures generalized linear models with normal (costs), and Poisson (utilization) distributions assessed differences in trends overall and separately for Blacks and Whites. With distress data for navigated survivors, we assessed high distress (score > 3), ≥1 distress cause (overall, by domain), and ≥ 1 assistance request by minority group. RESULTS: Beneficiaries were: 772 Black and 5350 White navigated, and 770 Black and 5348 White matched survivors. Impact was: i) costs: -$557.5 Blacks (p < .001), -$813.4 Whites (p < .001); ii) ER: Incidence Rate Ratio (IRR) 0.97 Blacks (NS), 0.93 Whites (p < .001); iii) hospitalizations: IRR 0.97 Blacks (NS), 0.91 Whites (p < .001). There was no significant difference in impact across minority groups. No significant differences were found in high distress (29% Black, 25.1% White), ≥1 distress cause (61.6% Black, 57.8% White), or ≥ 1 assistance request (64.5% Black, 59.1% White). Blacks were more likely to have ≥1 distress cause in the Practical domain. CONCLUSION: The PCCP may benefit both Black and White older cancer survivors. Programs should consider the proportion of older survivors with high distress, and the specific needs of minorities.

Unmet social support needs among older adults with cancer.

BACKGROUND: Adequate social support for older adults is necessary to maintain quality of life and reduce mortality and morbidity. However, little is known regarding the social support needs of older adults with cancer. The objective of the current study was to examine social support needs, specifically the unmet needs, among older adults with cancer. METHODS: Medicare beneficiaries (those aged ≥65 years) with cancer were identified from the University of Alabama at Birmingham Health System Cancer Community Network. Social support needs were assessed using a modified version of the Medical Outcomes Study Social Support Survey. The authors defined an "unmet need" if participants reported having some/a little/never availability of support and requiring support for that need. RESULTS: Of the 1460 participants in the current study, the average age was 74 years (standard deviation, 5.8 years). Approximately two-thirds of participants (986 participants; 67.5%) reported having at least 1 social support need, with the highest needs noted in the emotional (49.5%) and physical (47.4%) support subdomains. Of those individuals with a support need, approximately 45% had at least 1 unmet need, with the greatest percentages noted in the medical (39%) and informational (36%) subdomains. Multivariable analyses demonstrated that participants who were nonwhite, were divorced or never married, or had a high symptom burden were at greatest risk of having unmet social support needs across subdomains. CONCLUSIONS: In this population of older adults with cancer, the authors found high levels of unmet social support needs, particularly in the medical and informational support subdomains. Participants who were nonwhite, were divorced or never married, or had a high symptom burden were found to be at greatest risk of having unmet needs.

21st century research in urban WASH and health in sub-Saharan Africa: methods and outcomes in transition.

Tackling global inequalities in access to Water, Sanitation and Hygiene (WASH) remains an urgent issue - 58% of annual diarrhoeal deaths are caused by inadequate WASH provision. A global context of increasing urbanisation, changing demographics and health transitions demands an understanding and impact of WASH on a broad set of health outcomes. We examine the literature, in terms of health outcomes, considering WASH access and interventions in urban sub-Saharan Africa from 2000 to 2017. Our review of studies which evaluate the effectiveness of specific WASH interventions, reveals an emphasis of WASH research on acute communicable diseases, particularly diarrhoeal diseases. In contrast, chronic communicable and non-communicable health outcomes were notable gaps in the literature as well as a lack of focus on cross-cutting issues, such as ageing, well-being and gender equality. We recommend a broader focus of WASH research and interventions in urban Africa to better reflect the demographic and health transitions happening. Abbreviations: CBA: Controlled Before and After; GSD: Government Service Delivery; IWDSSD: International Drinking-Water, Supply and Sanitation Decade (IDWSSD); KAP: Knowledge, Attitudes and Practices; IBD: Irritable Bowel Diseases; MDG: Millennium Development Goals; NTD: Neglected Tropical Diseases; PSSD: Private Sector Service Delivery; SDG: Sustainable Development Goals; SSA: Sub Saharan Africa; SODIS: Solar Disinfection System; STH: Soil Transmitted Helminths; RCT: Randomised Control Trial; WASH: Water Sanitation and Hygiene; WHO: World Health Organization.