A Japanese retrospective study of non-tuberculous mycobacterial infection in children, adolescents, and young adult patients with hematologic-oncologic diseases.

Non-tuberculous mycobacterial infection (NTM) is rare in healthy children, with lymphadenitis being the most common presentation. Immunocompromised populations are known to be at high risk, but the clinical picture of NTM infection in pediatric hematology/oncology patients is unclear. In this nationwide retrospective analysis of patients under the age of 40 treated in Japanese pediatric hematology/oncology departments who developed NTM infection between January 2010 and December 2020, 36 patients (21 patients with hematopoietic stem cell transplantation (HSCT) and 15 nontransplant patients) were identified. Post-transplant patients were infected with NTM at 24 sites, including the lungs (n = 12), skin and soft tissues (n = 6), bloodstream (n = 4), and others (n = 2). Nine of twelve patients with pulmonary NTM infection had a history of pulmonary graft-versus-host disease (GVHD), and rapid-growing mycobacteria (RGM) were isolated from five of them. In nontransplant patients, the primary diseases were acute lymphoblastic leukemia (ALL; n = 5), inborn errors of immunity (IEI; n = 6), and others (n = 4). All cases of ALL had bloodstream infections with RGM, whereas all cases of IEI were infected with slow-growing mycobacteria (SGM). In summary, three typical clinical scenarios for pediatric hematology/oncology patients have been established: RGM-induced pulmonary disease in patients with pulmonary GVHD, RGM bloodstream infection in patients with ALL, and SGM infection in patients with IEI. Our findings suggest that NTM must be regarded as a pathogen for infections in these high-risk patients, especially those with pulmonary GVHD, who may require active screening for NTM.

In-Hospital Cancer Mortality Prediction by Multimodal Learning of Non-English Clinical Texts.

Predicting important outcomes in patients with complex medical conditions using multimodal electronic medical records remains challenge. We trained a machine learning model to predict the inpatient prognosis of cancer patients using EMR data with Japanese clinical text records, which has been considered difficult due to its high context. We confirmed high accuracy of the mortality prediction model using clinical text in addition to other clinical data, suggesting applicability of this method to cancer.