Promises, Pitfalls, and Clinical Applications of Artificial Intelligence in Pediatrics.

Artificial intelligence (AI) broadly describes a branch of computer science focused on developing machines capable of performing tasks typically associated with human intelligence. Those who connect AI with the world of science fiction may meet its growing rise with hesitancy or outright skepticism. However, AI is becoming increasingly pervasive in our society, from algorithms helping to sift through airline fares to substituting words in emails and SMS text messages based on user choices. Data collection is ongoing and is being leveraged by software platforms to analyze patterns and make predictions across multiple industries. Health care is gradually becoming part of this technological transformation, as advancements in computational power and storage converge with the rapid expansion of digitized medical information. Given the growing and inevitable integration of AI into health care systems, it is our viewpoint that pediatricians urgently require training and orientation to the uses, promises, and pitfalls of AI in medicine. AI is unlikely to solve the full array of complex challenges confronting pediatricians today; however, if used responsibly, it holds great potential to improve many aspects of care for providers, children, and families. Our aim in this viewpoint is to provide clinicians with a targeted introduction to the field of AI in pediatrics, including key promises, pitfalls, and clinical applications, so they can play a more active role in shaping the future impact of AI in medicine.

Primary Care Enhanced Access Services and the Association With Nonurgent Pediatric Emergency Department Utilization and Child Opportunity Index.

OBJECTIVES: This study aims to examine the association between primary care practice characteristics (enhanced access services) and practice-level rates of nonurgent emergency department (ED) visits using ED and practice-level data. Survey data suggest that enhanced access services within a child's primary care practice may be associated with reduced nonurgent ED visits. METHODS: We performed a cross-sectional analysis of nonurgent ED visits to a tertiary pediatric hospital in Western Pennsylvania with nearly 85,000 annual ED visits. We obtained patient encounter data of all nonurgent pediatric ED (PED) visits between January 2018 and December 2019. We identified the primary care provider at the time of the study period. For each of the 42 included offices, we determined the number of unique children in the office with a nonurgent PED visit, allowing us to determine the percentage of children in the practice with such a visit during the study period. We then stratified the 42 offices into low, intermediate, and high tertiles of nonurgent PED use. Using Kruskal-Wallis tests, logistic regression, and Pearson χ2 tests, we compared practice characteristics, enhanced access services, practice location Child Opportunity Index 2.0, and PED visit diagnoses across tertiles. RESULTS: We examined 52,459 nonurgent PED encounters by 33,209 unique patients across 42 outpatient offices. Primary care practices in the lowest ED visit tertile were more likely to have 4 or more evenings with office hours (36% vs 14%, P = 0.04), 4 or more evenings of weekday extended hours (43% vs 14%, P = 0.05), and at least 1 day of any weekend hours (86% vs 29%, P = 0.01), compared with practices in other tertiles. High PED use tertile offices were also associated with lower Child Opportunity Index scores. CONCLUSIONS: Primary care offices with higher nonurgent PED utilization had fewer enhanced access services and were located in neighborhood with fewer child-focused resources.

Large Language Models in Pediatric Education: Current Uses and Future Potential.

Generative artificial intelligence, especially large language models (LLMs), has the potential to affect every level of pediatric education and training. Demonstrating speed and adaptability, LLMs can aid educators, trainees, and practicing pediatricians with tasks such as enhancing curriculum design through the creation of cases, videos, and assessments; creating individualized study plans and providing real-time feedback for trainees; and supporting pediatricians by enhancing information searches, clinic efficiency, and bedside teaching. LLMs can refine patient education materials to address patients' specific needs. The current versions of LLMs sometimes provide "hallucinations" or incorrect information but are likely to improve. There are ethical concerns related to bias in the output of LLMs, the potential for plagiarism, and the possibility of the overuse of an online tool at the expense of in-person learning. The potential benefits of LLMs in pediatric education can outweigh the potential risks if employed judiciously by content experts who conscientiously review the output. All stakeholders must firmly establish rules and policies to provide rigorous guidance and assure the safe and proper use of this transformative tool in the care of the child. In this article, we outline the history, current uses, and challenges with generative artificial intelligence in pediatrics education. We provide examples of LLM output, including performance on a pediatrics examination guide and the creation of patient care instructions. Future directions to establish a safe and appropriate path for the use of LLMs will be discussed.

Artificial Intelligence and Objective Structured Clinical Examinations: Using ChatGPT to Revolutionize Clinical Skills Assessment in Medical Education.

This article examines the integration of OpenAI's Chat Generative Pre-trained Transformer (ChatGPT) into Objective Structured Clinical Examinations (OSCEs) for medical education. OSCEs, essential in evaluating medical trainees, are time and resource-intensive for educators and medical colleges. ChatGPT emerges as a solution, aiding educators in efficient OSCE preparation, including case development, standardized patient training, assessment methods, and grading rubrics. We explore ChatGPT's role in reducing trainee stress through simulated interactions of realistic practice scenarios and real-time trainee feedback. We also discuss the importance of validating ChatGPT outputs for medical accuracy and address compliance concerns. While highlighting ChatGPT's potential in reducing time and cost burdens for educators, we underscore the need for careful and informed application of Artificial Intelligence in medical education. Through examples, we outline ChatGPT's promising future in augmenting medical training and assessment, balancing technological innovation with educational integrity.

A randomized, embedded, pragmatic, Bayesian clinical trial examining clinical decision support for high flow nasal cannula management in children with bronchiolitis: design and statistical analysis plan.

BACKGROUND: High flow nasal cannula (HFNC) has been increasingly adopted in the past 2 decades as a mode of respiratory support for children hospitalized with bronchiolitis. The growing use of HFNC despite a paucity of high-quality data regarding the therapy's efficacy has led to concerns about overutilization. We developed an electronic health record (EHR) embedded, quality improvement (QI) oriented clinical trial to determine whether standardized management of HFNC weaning guided by clinical decision support (CDS) results in a reduction in the duration of HFNC compared to usual care for children with bronchiolitis. METHODS: The design and summary of the statistical analysis plan for the REspiratory SupporT for Efficient and cost-Effective Care (REST EEC; "rest easy") trial are presented. The investigators hypothesize that CDS-coupled, standardized HFNC weaning will reduce the duration of HFNC, the trial's primary endpoint, for children with bronchiolitis compared to usual care. Data supporting trial design and eventual analyses are collected from the EHR and other real world data sources using existing informatics infrastructure and QI data sources. The trial workflow, including randomization and deployment of the intervention, is embedded within the EHR of a large children's hospital using existing vendor features. Trial simulations indicate that by assuming a true hazard ratio effect size of 1.27, equivalent to a 6-h reduction in the median duration of HFNC, and enrolling a maximum of 350 children, there will be a > 0.75 probability of declaring superiority (interim analysis posterior probability of intervention effect > 0.99 or final analysis posterior probability of intervention effect > 0.9) and a > 0.85 probability of declaring superiority or the CDS intervention showing promise (final analysis posterior probability of intervention effect > 0.8). Iterative plan-do-study-act cycles are used to monitor the trial and provide targeted education to the workforce. DISCUSSION: Through incorporation of the trial into usual care workflows, relying on QI tools and resources to support trial conduct, and relying on Bayesian inference to determine whether the intervention is superior to usual care, REST EEC is a learning health system intervention that blends health system operations with active evidence generation to optimize the use of HFNC and associated patient outcomes. TRIAL REGISTRATION: ClinicalTrials.gov NCT05909566. Registered on June 18, 2023.

Prognostic clinical decision support for pneumonia in the emergency department: A randomized trial.

BACKGROUND: Hospitalization rates for childhood pneumonia vary widely. Risk-based clinical decision support (CDS) interventions may reduce unwarranted variation. METHODS: We conducted a pragmatic randomized trial in two US pediatric emergency departments (EDs) comparing electronic health record (EHR)-integrated prognostic CDS versus usual care for promoting appropriate ED disposition in children (<18 years) with pneumonia. Encounters were randomized 1:1 to usual care versus custom CDS featuring a validated pneumonia severity score predicting risk for severe in-hospital outcomes. Clinicians retained full decision-making authority. The primary outcome was inappropriate ED disposition, defined as early transition to lower- or higher-level care. Safety and implementation outcomes were also evaluated. RESULTS: The study enrolled 536 encounters (269 usual care and 267 CDS). Baseline characteristics were similar across arms. Inappropriate disposition occurred in 3% of usual care encounters and 2% of CDS encounters (adjusted odds ratio: 0.99, 95% confidence interval: [0.32, 2.95]) Length of stay was also similar and adverse safety outcomes were uncommon in both arms. The tool's custom user interface and content were viewed as strengths by surveyed clinicians (>70% satisfied). Implementation barriers include intrinsic (e.g., reaching the right person at the right time) and extrinsic factors (i.e., global pandemic). CONCLUSIONS: EHR-based prognostic CDS did not improve ED disposition decisions for children with pneumonia. Although the intervention's content was favorably received, low subject accrual and workflow integration problems likely limited effectiveness. Clinical Trials Registration: NCT06033079.

Pediatric Gastrointestinal Outcomes During the Post-Acute Phase of COVID-19: Findings from RECOVER Initiative from 29 Hospitals in the US.

Importance: The profile of gastrointestinal (GI) outcomes that may affect children in post-acute and chronic phases of COVID-19 remains unclear. Objective: To investigate the risks of GI symptoms and disorders during the post-acute phase (28 days to 179 days after SARS-CoV-2 infection) and the chronic phase (180 days to 729 days after SARS-CoV-2 infection) in the pediatric population. Design: We used a retrospective cohort design from March 2020 to Sept 2023. Setting: twenty-nine healthcare institutions. Participants: A total of 413,455 patients aged not above 18 with SARS-CoV-2 infection and 1,163,478 patients without SARS-CoV-2 infection. Exposures: Documented SARS-CoV-2 infection, including positive polymerase chain reaction (PCR), serology, or antigen tests for SARS-CoV-2, or diagnoses of COVID-19 and COVID-related conditions. Main Outcomes and Measures: Prespecified GI symptoms and disorders during two intervals: post-acute phase and chronic phase following the documented SARS-CoV-2 infection. The adjusted risk ratio (aRR) was determined using a stratified Poisson regression model, with strata computed based on the propensity score. Results: Our cohort comprised 1,576,933 patients, with females representing 48.0% of the sample. The analysis revealed that children with SARS-CoV-2 infection had an increased risk of developing at least one GI symptom or disorder in both the post-acute (8.64% vs. 6.85%; aRR 1.25, 95% CI 1.24-1.27) and chronic phases (12.60% vs. 9.47%; aRR 1.28, 95% CI 1.26-1.30) compared to uninfected peers. Specifically, the risk of abdominal pain was higher in COVID-19 positive patients during the post-acute phase (2.54% vs. 2.06%; aRR 1.14, 95% CI 1.11-1.17) and chronic phase (4.57% vs. 3.40%; aRR 1.24, 95% CI 1.22-1.27). Conclusions and Relevance: In the post-acute phase or chronic phase of COVID-19, the risk of GI symptoms and disorders was increased for COVID-positive patients in the pediatric population.

User-Centered Design and Implementation of an Interoperable FHIR Application for Pediatric Pneumonia Prognostication in a Randomized Trial.

OBJECTIVES: To support a pragmatic, electronic health record (EHR)-based randomized controlled trial, we applied user-centered design (UCD) principles, evidence-based risk communication strategies, and interoperable software architecture to design, test, and deploy a prognostic tool for children in emergency departments (EDs) with pneumonia. METHODS: Risk for severe in-hospital outcomes was estimated using a validated ordinal logistic regression model to classify pneumonia severity. To render the results usable for ED clinicians, we created an integrated SMART on Fast Healthcare Interoperability Resources (FHIR) web application built for interoperable use in two pediatric EDs using different EHR vendors: Epic and Cerner. We followed a UCD framework, including problem analysis and user research, conceptual design and early prototyping, user interface development, formative evaluation, and postdeployment summative evaluation. RESULTS: Problem analysis and user research from 39 clinicians and nurses revealed user preferences for risk aversion, accessibility, and timing of risk communication. Early prototyping and iterative design incorporated evidence-based design principles, including numeracy, risk framing, and best-practice visualization techniques. After rigorous unit and end-to-end testing, the application was successfully deployed in both EDs, which facilitated enrollment, randomization, model visualization, data capture, and reporting for trial purposes. CONCLUSION: The successful implementation of a custom application for pneumonia prognosis and clinical trial support in two health systems on different EHRs demonstrates the importance of UCD, adherence to modern clinical data standards, and rigorous testing. Key lessons included the need for understanding users' real-world needs, regular knowledge management, application maintenance, and the recognition that FHIR applications require careful configuration for interoperability.