Age-Dependent Variations in Kawasaki Disease Incidence in Japan.

Importance: The etiology of Kawasaki disease (KD) remains elusive, with immunologic and epidemiologic data suggesting different triggers in individuals who are genetically susceptible. KD remains the most common cause of acquired heart disease in pediatric patients, and Japan is the country of highest incidence, with an increasing number of cases. Objective: To investigate whether an analysis of the epidemiologic KD record in Japan stratified by age and prefecture (subregion) may yield new clues regarding mechanisms of exposure to etiologic agents associated with KD. Design, Setting, and Participants: This cross-sectional study was conducted using a dataset of patients with KD with detailed information on location and age at onset created through nationwide surveys of hospitals caring for pediatric patients with KD throughout Japan. Pediatric patients hospitalized in Japan for KD from 1970 to 2020 were included. Data were analyzed from January 2022 to January 2024. Exposure: Pediatric patients with KD. Main Outcomes and Measures: The KD dataset was analyzed by patient age (infants [aged <6 months], toddlers [aged 6 to <24 months], children aged 2 years [aged 24 to <36 months], and children and adolescents aged 3 years or older [aged ≥36 months]), with investigations of seasonal cycles, interannual variations, and correlations across regions. Results: Among 422 528 pediatric patients (243 803 males [57.7%] and 178 732 females [42.3%]; median [IQR] age, 23.69 [11.96-42.65] months), infants, toddlers, and patients aged 3 years or older exhibited different rates of increase in KD incidence, seasonality, and degrees of coherence of seasonality across prefectures. Although the mean (SD) incidence of KD among infants remained relatively stable over the past 30 years compared with older patients (1.00 [0.07] in 1987-1992 to 2.05 [0.11] in 2011-2016), the mean (SD) incidence rate for children and adolescents aged 3 years or older increased 5.2-fold, from 1.00 (0.08) in 1987 to 1992 to 5.17 (0.46) in 2014 to 2019. Patients aged 3 years or older saw a reduction in mean (SD) incidence, from peaks of 5.71 (0.01) in October 2014 through June 2015 and July 2018 through March 2019 to 4.69 (0.11) in 2016 to 2017 (17.8% reduction) not seen in younger children. The seasonal cycle varied by age group; for example, mean (SD) incidence peaked in July and August (5.63 [0.07] cases/100 000 individuals) for infants and in December and January (4.67 [0.13] cases/100 000 individuals) for toddlers. Mean (SD) incidence changed dramatically for toddlers beginning in the early 2010s; for example, the normalized mean (SD) incidence among toddlers for October was 0.74 (0.03) in 1992 to 1995 and 1.10 (0.01) in 2016 to 2019. Across Japan, the seasonal cycle of KD incidence of older children and adolescents exhibited mean (SD) correlation coefficients between prefectures as high as 0.78 (0.14) for prefecture 14 among patients aged 3 years or older, while that of infants was much less (highest mean [SD] correlation coefficient, 0.43 [0.23]). Conclusions and Relevance: This study found distinct temporal signatures and changing spatial consistency of KD incidence across age groups, suggesting different age-related mechanisms of exposure. Some results suggested that social factors may modulate exposure to etiologic agents of KD; however, the increase in KD incidence in older children coupled with the correlation across prefectures of KD incidence suggest that the intensity of an environmental exposure that triggers KD in this age group may have increased over time.

Fast and Efficient Root Phenotyping via Pose Estimation.

Image segmentation is commonly used to estimate the location and shape of plants and their external structures. Segmentation masks are then used to localize landmarks of interest and compute other geometric features that correspond to the plant's phenotype. Despite its prevalence, segmentation-based approaches are laborious (requiring extensive annotation to train) and error-prone (derived geometric features are sensitive to instance mask integrity). Here, we present a segmentation-free approach that leverages deep learning-based landmark detection and grouping, also known as pose estimation. We use a tool originally developed for animal motion capture called SLEAP (Social LEAP Estimates Animal Poses) to automate the detection of distinct morphological landmarks on plant roots. Using a gel cylinder imaging system across multiple species, we show that our approach can reliably and efficiently recover root system topology at high accuracy, few annotated samples, and faster speed than segmentation-based approaches. In order to make use of this landmark-based representation for root phenotyping, we developed a Python library (sleap-roots) for trait extraction directly comparable to existing segmentation-based analysis software. We show that pose-derived root traits are highly accurate and can be used for common downstream tasks including genotype classification and unsupervised trait mapping. Altogether, this work establishes the validity and advantages of pose estimation-based plant phenotyping. To facilitate adoption of this easy-to-use tool and to encourage further development, we make sleap-roots, all training data, models, and trait extraction code available at: https://github.com/talmolab/sleap-roots and https://osf.io/k7j9g/.

Fast and efficient root phenotyping via pose estimation.

Image segmentation is commonly used to estimate the location and shape of plants and their external structures. Segmentation masks are then used to localize landmarks of interest and compute other geometric features that correspond to the plant's phenotype. Despite its prevalence, segmentation-based approaches are laborious (requiring extensive annotation to train), and error-prone (derived geometric features are sensitive to instance mask integrity). Here we present a segmentation-free approach which leverages deep learning-based landmark detection and grouping, also known as pose estimation. We use a tool originally developed for animal motion capture called SLEAP (Social LEAP Estimates Animal Poses) to automate the detection of distinct morphological landmarks on plant roots. Using a gel cylinder imaging system across multiple species, we show that our approach can reliably and efficiently recover root system topology at high accuracy, few annotated samples, and faster speed than segmentation-based approaches. In order to make use of this landmark-based representation for root phenotyping, we developed a Python library (sleap-roots) for trait extraction directly comparable to existing segmentation-based analysis software. We show that landmark-derived root traits are highly accurate and can be used for common downstream tasks including genotype classification and unsupervised trait mapping. Altogether, this work establishes the validity and advantages of pose estimation-based plant phenotyping. To facilitate adoption of this easy-to-use tool and to encourage further development, we make sleap-roots, all training data, models, and trait extraction code available at: https://github.com/talmolab/sleap-roots and https://osf.io/k7j9g/.