Brian R. Englum

Background: We analyzed trends in survival of children and adolescents diagnosed with cancer in the United States from 1975 to 2022 using the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) 8 and 17 databases. Our goal was to provide an updated overview of outcomes and highlight disparities to guide future research. Methods: Data were obtained from the SEER 8 and 17 registries for patients diagnosed at ages 0-19 years. Overall survival was stratified by race and ethnicity, diagnosis, and population density. Temporal trends in 5-year overall survival were evaluated using the National Cancer Institute Joinpoint Regression Program. Kaplan–Meier survival curves with log-rank tests assessed survival differences across subgroups. Multivariable Cox proportional hazards models evaluated overall survival, adjusted for age, decade of diagnosis, and race and ethnicity. Results: A total of 52 165 patients were diagnosed with cancer in SEER 8 and 100 110 in SEER 17. Five-year overall survival for all cancers improved from 60% (95% CI = 56% to 64%) in 1975 to 90% (95% CI = 88% to 91%) in 2017. Survival gains were greatest in acute lymphoblastic leukemia, with more modest improvements in solid tumors. Mortality beyond 20 years persisted for some cancers, including Hodgkin lymphoma. Survival differed statistically by race and ethnicity (P < .001), with non-Hispanic Black patients experiencing the lowest 5-year overall survival across multiple cancers, even after adjusting for age and decade of diagnosis. Conclusions: Survival for childhood and adolescent cancers has markedly improved over the past 5 decades, yet progress remains uneven across cancer types and race and ethnicity. Understanding drivers of disparities is essential to achieve equitable outcomes for all young cancer patients.

Background: Over 90% of US children with cancer are treated at Children's Oncology Group (COG) centers, which seek to maximize enrollment in therapeutic and biobanking studies. Rare cancers have demonstrated lower than expected COG enrollment. We evaluated trends in COG rare cancer enrollment compared to US incidence from Surveillance, Epidemiology, and End Results (SEER) registries, examining the impact of COG therapeutic trials and Project:EveryChild, a cancer biobank/registry. Procedure: COG and SEER data from 2002 to 2020 were queried for US patients younger than 18 years old with adrenocortical carcinoma (ACC), nasopharyngeal carcinoma (NPC), retinoblastoma (RB), thyroid carcinoma, and melanoma. We compared demographic data between COG and SEER, extrapolating incidence for each cancer to analyze trends in COG enrollment. Results: Patient characteristics, including age, sex, and race, were similar between COG ( n = 2184) and SEER ( n = 5514). COG enrollment for rare cancers remained low (11%). Initiating Project:EveryChild did not increase enrollment (12% pre‐ vs. 8% post‐Project:EveryChild; p < 0.01). For cancers with available therapeutic trials (ACC, NPC, and RB), COG enrollment was higher during trial accrual (40%) than when no trial was open (12%; p < 0.01). Patient geography and income did not appear as barriers to COG enrollment. Conclusions: Although children with rare cancers enrolled in COG studies reflect the US population, enrollment in the COG registry/biospecimen repository continues to be limited in the absence of therapeutic trials, impacting data and biospecimens available to inform therapeutic trial development. Expansion of therapeutic trials or free molecular testing through the Molecular Characterization Initiative may increase data and biospecimens for these rare cancers.

The coronavirus disease-2019 (COVID-19) pandemic caused disruptions in cancer screening and diagnosis, including colorectal cancer. Given disparities in ca

Rapid and accurate identification of Venous thromboembolism (VTE), a severe cardiovascular condition including deep vein thrombosis (DVT) and pulmonary embolism (PE), is important for effective treatment. Leveraging Natural Language Processing (NLP) on radiology reports, automated methods have shown promising advancements in identifying VTE events from retrospective data cohorts or aiding clinical experts in identifying VTE events from radiology reports. However, effectively training Deep Learning (DL) and the NLP models is challenging due to limited labeled medical text data, the complexity and heterogeneity of radiology reports, and data imbalance. This study proposes novel method combinations of DL methods, along with data augmentation, adaptive pre-trained NLP model selection, and a clinical expert NLP rule-based classifier, to improve the accuracy of VTE identification in unstructured (free-text) radiology reports. Our experimental results demonstrate the model's efficacy, achieving an impressive 97\% accuracy and 97\% F1 score in predicting DVT, and an outstanding 98.3\% accuracy and 98.4\% F1 score in predicting PE. These findings emphasize the model's robustness and its potential to significantly contribute to VTE research.