Background: Whether GPT-4, the conversational artificial intelligence, can accurately diagnose and triage health conditions and whether it presents racial and ethnic biases in its decisions remain unclear. Objective: We aim to assess the accuracy of GPT-4 in the diagnosis and triage of health conditions and whether its performance varies by patient race and ethnicity. Methods: We compared the performance of GPT-4 and physicians, using 45 typical clinical vignettes, each with a correct diagnosis and triage level, in February and March 2023. For each of the 45 clinical vignettes, GPT-4 and 3 board-certified physicians provided the most likely primary diagnosis and triage level (emergency, nonemergency, or self-care). Independent reviewers evaluated the diagnoses as “correct” or “incorrect.” Physician diagnosis was defined as the consensus of the 3 physicians. We evaluated whether the performance of GPT-4 varies by patient race and ethnicity, by adding the information on patient race and ethnicity to the clinical vignettes. Results: The accuracy of diagnosis was comparable between GPT-4 and physicians (the percentage of correct diagnosis was 97.8% (44/45; 95% CI 88.2%-99.9%) for GPT-4 and 91.1% (41/45; 95% CI 78.8%-97.5%) for physicians; P=.38). GPT-4 provided appropriate reasoning for 97.8% (44/45) of the vignettes. The appropriateness of triage was comparable between GPT-4 and physicians (GPT-4: 30/45, 66.7%; 95% CI 51.0%-80.0%; physicians: 30/45, 66.7%; 95% CI 51.0%-80.0%; P=.99). The performance of GPT-4 in diagnosing health conditions did not vary among different races and ethnicities (Black, White, Asian, and Hispanic), with an accuracy of 100% (95% CI 78.2%-100%). P values, compared to the GPT-4 output without incorporating race and ethnicity information, were all .99. The accuracy of triage was not significantly different even if patients’ race and ethnicity information was added. The accuracy of triage was 62.2% (95% CI 46.5%-76.2%; P=.50) for Black patients; 66.7% (95% CI 51.0%-80.0%; P=.99) for White patients; 66.7% (95% CI 51.0%-80.0%; P=.99) for Asian patients, and 62.2% (95% CI 46.5%-76.2%; P=.69) for Hispanic patients. P values were calculated by comparing the outputs with and without conditioning on race and ethnicity. Conclusions: GPT-4’s ability to diagnose and triage typical clinical vignettes was comparable to that of board-certified physicians. The performance of GPT-4 did not vary by patient race and ethnicity. These findings should be informative for health systems looking to introduce conversational artificial intelligence to improve the efficiency of patient diagnosis and triage.

Sepsis is caused by the body’s dysregulated response to infection, which can lead to multiorgan injury and death. Patients with sepsis may develop acute cardiac dysfunction, termed septic cardiomyopathy, which is a global but reversible dysfunction of both sides of the heart. This narrative review discusses the mechanistic changes in the heart during septic cardiomyopathy, its diagnosis, existing treatment options regarding severity and course, and emerging treatment approaches. Although no standardized definition for septic cardiomyopathy exists, it is described as a reversible myocardial dysfunction that typically resolves within 7 to 10 days. Septic cardiomyopathy is often diagnosed based on electrocardiography, cardiac magnetic resonance imaging, biomarkers, and direct invasive and noninvasive measures of cardiac output. Presently, the treatment of septic cardiomyopathy is similar to that of sepsis, primarily focusing on acute interventions. Treatments for cardiomyopathy often include angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, and diuretics. However, because of profound hypotension in sepsis, many cardiomyopathy treatments are contraindicated in patients with septic cardiomyopathy. Substantial efforts have been made to study the pathophysiological mechanisms and diagnostic options; however, the lack of a uniform definition for septic cardiomyopathy is challenging for physicians when considering treatments. Another challenge for physicians is that the treatment for septic cardiomyopathy has only focused on acute intervention, whereas the treatment for other cardiomyopathies has been provided on a long-term basis. A better understanding of the underlying mechanisms of septic cardiomyopathy may contribute to the development of a unified definition of the condition and novel treatment options.

We retrospectively analyzed data from the Medical Information Mart for Intensive Care-III critical care database to determine whether visually-assessed right ventricular (RV) dysfunction was associated with clinical outcomes in septic shock patients. Associations between visually-assessed RV dysfunction by echocardiography and in-hospital mortality, lethal arrhythmia, and hemodynamic indicators to determine the prognostic value of RV dysfunction in patients with septic shock were analyzed. Propensity score analysis showed RV dysfunction was associated with increased risk of in-hospital death in patients with septic shock (adjusted odds ratio [OR] 2.15; 95% confidence interval [CI] 1.99–2.32;P < 0.001). In multivariate logistic regression analysis, RV dysfunction was associated with in-hospital death (OR 2.19; 95% CI 1.91–2.53;P < 0.001), lethal arrhythmia (OR 2.19; 95% CI 1.34–3.57;P < 0.001), and tendency for increased blood lactate levels (OR 1.31; 95% CI 1.14–1.50;P < 0.001) independent of left ventricular (LV) dysfunction. RV dysfunction was associated with lower cardiac output, pulmonary artery pressure index, and RV stroke work index. In patients with septic shock, visually-assessed RV dysfunction was associated with in-hospital mortality, lethal arrhythmia, and circulatory insufficiency independent of LV dysfunction. Visual assessment of RV dysfunction using echocardiography might help to identify the short-term prognosis of patients with septic shock by reflecting hemodynamic status.

Acute kidney injury (AKI) is a major complication of sepsis that induces acid-base imbalances. While creatinine levels are the only indicator for assessing the prognosis of AKI, prognostic importance of metabolic acidosis is unknown. We conducted a retrospective observational study by analyzing a large China-based pediatric critical care database from 2010 to 2018. Participants were critically ill children with AKI admitted to intensive care units (ICUs). The study included 1505 children admitted to ICUs with AKI, including 827 males and 678 females. The median age at ICU admission was 22 months (interquartile range 7–65). After a median follow-up of 10.87 days, 4.3% (65 patients) died. After adjusting for confounding factors, hyperlactatemia, low pH, and low bicarbonate levels were independently associated with 28-day mortality (respective odds ratio: 3.06, 2.77, 2.09; p values: <0.01, <0.01, <0.01). The infection had no interaction with the three parameters. The AKI stage negatively interacted with bicarbonate and pH but not lactate. The current study shows that among children with AKI, hyperlactatemia, low pH, and hypobicarbonatemia are associated with 28-day mortality.

Studies over recent years have redeveloped our understanding of uremic cardiomyopathy, defined as left ventricular hypertrophy, congestive heart failure, and associated cardiac hypertrophy plus other abnormalities that result from chronic kidney disease and are often the cause of death in affected patients. Definitions of uremic cardiomyopathy have conflicted and overlapped over the decades, complicating the body of published evidence, and making comparison difficult. New and continuing research into potential risk factors, including uremic toxins, anemia, hypervolemia, oxidative stress, inflammation, and insulin resistance, indicates the increasing interest in illuminating the pathways that lead to UC and thereby identifying potential targets for intervention. Indeed, our developing understanding of the mechanisms of UC has opened new frontiers in research, promising novel approaches to diagnosis, prognosis, treatment, and management. This educational review highlights advances in the field of uremic cardiomyopathy and how they may become applicable in practice by clinicians. Pathways to optimal treatment with current modalities (with hemodialysis and angiotensin-converting enzyme inhibitors) will be described, along with proposed steps to be taken in research to allow evidence-based integration of developing investigational therapies.

The management of refractory septic shock remains a major challenge in critical care and its early indicators are not fully understood. We hypothesized that the maximum norepinephrine dosage within 24 hours of intensive care unit (ICU) admission may be a useful indicator of early mortality in patients with septic shock. In this retrospective single-center observational study, patients with septic shock admitted to the emergency ICU of an academic medical center between April 2011 and March 2017 were included. Individuals with cardiac arrest and those with do-not-resuscitate orders before admission were excluded. We analyzed if the maximum norepinephrine dosage within 24 hours of ICU admission (MD24) was associated with 7-day mortality. Among 152 patients with septic shock, 20 (15%) did not survive by day 7. The receiver operating characteristic curve analysis for predicting 7-day mortality revealed a cutoff of MD24 of 0.6 μg/kg/min (sensitivity 47%, specificity 93%). In the multivariable regression analysis, a higher MD24 was significantly associated with 7-day mortality (odds ratio: 7.20; 95% confidence interval [CI]: 2.02-25.7; P = .002) but not with 30-day mortality. Using the inverse probability of treatment weighting method in a propensity scoring analysis, a higher MD24 was significantly associated with 7-day (hazard ratio [HR]: 8.9; 95% CI: 3.2-25.0; P < .001) and 30-day mortality (HR: 2.7; 95% CI: 1.2-5.8; P = .012). An MD24 ≥0.6 μg/kg/min was significantly associated with 7-day mortality in patients with septic shock and may therefore be a useful indicator of refractory septic shock.