Background: Large language model (LLM)–based chatbots are evolving at an unprecedented pace with the release of ChatGPT, specifically GPT-3.5, and its successor, GPT-4. Their capabilities in general-purpose tasks and language generation have advanced to the point of performing excellently on various educational examination benchmarks, including medical knowledge tests. Comparing the performance of these 2 LLM models to that of Family Medicine residents on a multiple-choice medical knowledge test can provide insights into their potential as medical education tools. Objective: This study aimed to quantitatively and qualitatively compare the performance of GPT-3.5, GPT-4, and Family Medicine residents in a multiple-choice medical knowledge test appropriate for the level of a Family Medicine resident. Methods: An official University of Toronto Department of Family and Community Medicine Progress Test consisting of multiple-choice questions was inputted into GPT-3.5 and GPT-4. The artificial intelligence chatbot’s responses were manually reviewed to determine the selected answer, response length, response time, provision of a rationale for the outputted response, and the root cause of all incorrect responses (classified into arithmetic, logical, and information errors). The performance of the artificial intelligence chatbots were compared against a cohort of Family Medicine residents who concurrently attempted the test. Results: GPT-4 performed significantly better compared to GPT-3.5 (difference 25.0%, 95% CI 16.3%-32.8%; McNemar test: P<.001); it correctly answered 89/108 (82.4%) questions, while GPT-3.5 answered 62/108 (57.4%) questions correctly. Further, GPT-4 scored higher across all 11 categories of Family Medicine knowledge. In 86.1% (n=93) of the responses, GPT-4 provided a rationale for why other multiple-choice options were not chosen compared to the 16.7% (n=18) achieved by GPT-3.5. Qualitatively, for both GPT-3.5 and GPT-4 responses, logical errors were the most common, while arithmetic errors were the least common. The average performance of Family Medicine residents was 56.9% (95% CI 56.2%-57.6%). The performance of GPT-3.5 was similar to that of the average Family Medicine resident (P=.16), while the performance of GPT-4 exceeded that of the top-performing Family Medicine resident (P<.001). Conclusions: GPT-4 significantly outperforms both GPT-3.5 and Family Medicine residents on a multiple-choice medical knowledge test designed for Family Medicine residents. GPT-4 provides a logical rationale for its response choice, ruling out other answer choices efficiently and with concise justification. Its high degree of accuracy and advanced reasoning capabilities facilitate its potential applications in medical education, including the creation of exam questions and scenarios as well as serving as a resource for medical knowledge or information on community services.

Vaccinationis a critical tool in the prevention of COVID‐19 infection for individuals and for communities. The mRNA vaccines contain polyethylene glycol (PEG) as a stabilizer. Currently, in North America, only the BNT162b2 (Pfizer‐BioNTech) mRNA vaccine is approved for individuals aged 12–17. Most patients treated with contemporary regimens for acute lymphoblastic leukemia receive PEG‐asparaginase (PEG‐ASNase) and 10%–30% will develop allergic reactions. Optimizing access and safety for vaccine administration for these patients is critical. This report describes a process developed to support COVID vaccination in a cohort of adolescents and young adults with a history of PEG‐ASNase allergy.

Understanding the imaging spectrum of non-Hodgkin lymphoma in children, adolescents, and young adults is critical for radiologists to support clinical decision making by treating physicians and oth... In children, adolescents, and young adults (CAYA), non-Hodgkin lymphoma (NHL) is characterized by various age-related dissimilarities in tumor aggressiveness, prevailing pathologic subtypes, and im...

Introduction: Children with Down Syndrome (DS) have a 20-fold increased incidence of acute lymphoblastic leukemia (ALL) and a 70% higher risk of relapse than children with ALL and no DS. Treatment for DS children with relapsed ALL is challenging due to the high treatment-related mortality associated with both intensive chemotherapy and hematopoietic stem cell transplantation (HSCT) as well as high rates of second relapse. Outcomes for relapsed ALL in DS are dismal, with recent studies citing an event-free survival (EFS) as low as 17%. Alternative treatment approaches are needed for this patient population. Treatment with anti-CD19 chimeric antigen receptor T-cells (CARTs) have shown high remission rates in patients with multiple relapsed/refractory ALL with an acceptable safety profile. CARTs are generally not indicated for first relapse of ALL; DS-specific indications for CARTs do not exist. In Ontario, due to the specific susceptibility of children with DS-ALL, CARTs were made available to this population upon first relapse. We describe the population-based Canadian experience using CARTs for first relapse of DS-ALL. Methods: Efficacy and safety data on all children referred for CARTs at the Hospital for Sick Children, in Toronto, Canada were retrospectively collected in a central database and validated by treating clinicians. Here we report all patients with DS suffering a first relapse of ALL at age 36 months from diagnosis and one patient with early relapse. Mononuclear cells were successfully collected for all patients on the first attempt. Patients were bridged from collection to infusion with low intensity chemotherapy regimens. Tisagenlecleucel was successfully manufactured for all patients and infused at a median of 69 days (range 56-101) from collection of T cells. Four patients developed CRS: two patients grade 1, one patient grade 2, and one patient grade 4 requiring ICU care and treatment with tocilizumab and steroid. All patients achieved minimal residual disease (MRD) negative remission one month after infusion, with B-cell aplasia. Two patients showed early (< 6months post infusion) B-cell recovery consistent with CAR T-cell loss; both received off-label reinfusions of tisagenlecleucel and again achieved B-cell aplasia. One patient had a seizure with the first Fludarabine dose of lymphodepleting (LD) chemotherapy for re-infusion and was found to have elevated intracranial pressure. After treatment with Topiramate and anti-seizure prophylaxis the patient tolerated LD chemotherapy without further neurological events. No patient received further ALL therapy; none went on to HSCT. One patient suffered a CD19-negative relapse 7 months post infusion and was transitioned to palliative-intent treatment. Two patients developed prolonged neutropenia with one patient showing recovery (ANC>1x10E9/l) at 5 months post infusion and the patient who had relapsed disease remained neutropenic until relapse. Infectious complications in the first 6 months post infusion were rare. One patient, who was not neutropenic, required two hospital admissions for line related sepsis. With a median follow-up time of 519 days (range 197-785), the 1-year estimated EFS and overall survival were both 80% (95 ^th confidence interval 52-100%; Figure 1). Conclusions: Tisagenlecleucel was safe and effective in a small cohort of children with DS and first relapse of CD19+ ALL, with achievement of long-term remissions without further therapy, and an EFS superior to that seen in historical cohorts. Rate of acute complications were not increased compared with reported outcomes in patients without DS. Further evaluation of tisagenlecleucel in this setting is warranted.