Background: Translocation morphology renal cell carcinoma (tRCC) accounts for nearly half of all pediatric RCC cases. Biological study AREN14B4-Q aimed to characterize the molecular landscape of tRCC using samples acquired from patients enrolled in the Children's Oncology Group Risk Classification and Biobanking study AREN03B2. Methods: From 2006 to 2014, patients <30 yr old with renal tumors were prospectively enrolled in AREN03B2, a Central IRB-approved biobanking study. All pediatric RCC cases underwent a detailed central pathology review and molecular diagnostics to accurately classify RCC subtypes. Samples with confirmed tRCC and appropriate informed consent were identified with adequate tissue for RNA and DNA extraction, along with germline DNA, for whole-genome sequencing (WGS), RNA sequencing, and DNA methylation analyses. Results: From 41 patients, high-quality samples allowed for 18 tumors and non-tumor DNA to be analyzed via WGS, 19 via DNA methylation, and 36 RNA samples via transcriptome sequencing. Consistent with and extending clinical cytogenetic findings, WGS and fusion transcript analyses confirmed very few additional mutations beyond the tRCC translocation. No recurrent genomic copy number gains/losses were found. RNA and WGS analyses enabled sub-classification of tRCC, closely aligning with the different TFE3 fusion partners. DNA methylation analyses demonstrated less tRCC sub-stratification compared with RNA analyses. Pathways activated in tRCC were involved in epithelial differentiation, extracellular matrix organization, apoptosis, immune regulation, signal transduction, and angiogenesis. Conclusions: Arrested epithelial differentiation is the overarching driver in tRCC and is strongly correlated with the specific subclasses of fusion transcript generated by the genetic translocation TFE fusion partner. Negative regulation of apoptosis, increased M2 macrophage expression, and enhanced angiogenesis also appear to be functional features of tRCCs, as are increased expression of matrix metalloproteinases, PI3K-AKT/mTOR/MAPK signaling, and mitochondrial metabolism, highlighting potential therapeutic options beyond direct targeting of the oncogenic driver fusions.

Abstract Single-cell and spatial omics offer unprecedented opportunities to decipher the mechanisms of disease, however, this process requires teams of experts, iterative trial-and-error and reasoning across modalities. Here we present LungChat ( https://chat.lungmap.net ), a conversational system for integrated multi-omics analysis and biomedical discovery, deployed as a hierarchical multi-agent architecture in which a supervisor decomposes natural-language questions into parallel, tool-grounded tasks spanning single-cell and spatial analyses, literature and clinical-trial synthesis, and drug repurposing. To predict new therapeutics, LungChat implements Direction-Aware Repurposing and Targeting (DART) to distinguish perturbations that reverse disease transcriptional programs from those that reinforce them, at the cell-type level, for safety prediction. Controlled architecture ablations showed that hierarchical orchestration improved grounded abstention and token efficiency and preserved strong performance on complex multi-step tasks. In pulmonary disease case studies, LungChat independently prioritized saracatinib for IPF through drug-connectivity screening, followed by DART-based cell-type analysis; the same compound has been evaluated in the STOP-IPF clinical trial ( NCT04598919 ). The system also recovered fluticasone propionate, an established COPD therapy, through a single orchestrated analysis. This tissue-agnostic system provides a blueprint for verifiable agentic AI systems that support reproducible scientific discovery.

Single-cell spatial transcriptomics can provide subcellular resolution for a deep understanding of molecular mechanisms. However, accurate segmentation and annotation remain a major challenge that limits downstream analysis. Current machine learning methods heavily rely on nuclei or cell body staining, resulting in the significant loss of both transcriptome depth and the limited ability to learn spatial colocalization patterns. Here, we propose Bering, a graph deep learning model that leverages transcript colocalization relationships for joint noise-aware cell segmentation and molecular annotation in 2D and 3D spatial transcriptomics data. To evaluate performance, we benchmark Bering with state-of-the-art methods and observe better cell segmentation accuracies and more detected transcripts across technologies and tissues. To streamline segmentation processes, we construct expansive pre-trained models, which yield high segmentation accuracy in new data through transfer learning and self-distillation. These improved capabilities enable Bering to enhance cell annotations for the rapidly expanding field of spatial omics.

BACKGROUND:Studies have demonstrated that patients with myocarditis may have a higher burden of cardiomyopathy-associated genetic variants than the general population. However, data on children are limited. We compared the prevalence of rare predicted-damaging variants and clinically pathogenic variants in children with dilated cardiomyopathy (DCM) secondary to myocarditis with that in children with DCM alone and in heart-healthy controls. METHODS:Children with DCM secondary to myocarditis and children with DCM alone who underwent exome sequencing as part of a prior cross-sectional study were identified in the Pediatric Cardiomyopathy Registry, a large multicenter registry of children with cardiomyopathy. Controls from the Indiana University Biobank were matched 4:1 with myocarditis cases on genomic similarity. Rare predicted-damaging variants in cardiomyopathy-associated genes were identified using a bioinformatics approach. Clinical guidelines were used to determine clinical pathogenicity. The prevalence of variants was compared across the 3 groups. RESULTS:There were 32 patients with DCM secondary to myocarditis. The prevalence of rare predicted-damaging variants was 34.4% (11/32 [95% CI, 18.6%-53.2%]) in cases compared with 6.3% (8/128 [95% CI, 2.7%-11.9%]) in controls (P<0.001). Clinical review indicated all rare predicted-damaging variants in cases were pathogenic (1/12), likely pathogenic (3/12), or variants of uncertain significance (8/12), whereas most variants in controls were benign (2/8) or likely benign (4/8). The prevalence of pathogenic/likely pathogenic variants in cases was 12.5% (95% CI, 3.5%-29.0%) compared with 0% (95% CI, 0%-2.3%) in controls (P<0.01). Rare predicted-damaging and clinically pathogenic/likely pathogenic variant prevalence was not significantly different in children with DCM secondary to myocarditis and DCM without myocarditis (P=0.17 and P=1.00, respectively). CONCLUSIONS:Children with DCM secondary to myocarditis had a higher burden of variants in cardiomyopathy-associated genes than that of heart-healthy controls. Larger studies will be needed to determine the utility of routine genetic testing in this population.

Hepatoblastoma (HB) is the most common pediatric liver malignancy. However, its cellular origin and molecular drivers remain poorly defined. Using single-nuclear RNA sequencing (snRNA-seq), we identified a proliferative, hepatocyte-derived tumor cell population (cycling HepT) enriched for Enhancer of Zeste Homolog 2 (EZH2) expression, particularly in the aggressive embryonal subtype. Integrative genomic and transcriptomic profiling confirmed EZH2 overexpression. Disruption of the PRC2 complex was evident through mislocalization and reduced expression of SUZ12, a core component. EZH2 overexpression correlated with upregulation of mitotic regulators such as AURKB and Ki67 in human HB gene expression analysis as compared to background liver. Targeted sequencing identified variants of uncertain significance in EZH2 and SUZ12 in 11 of 11 patient tumors. Pharmacologic inhibition of EZH2 with EPZ-6438 reduced proliferation and sensitized HB cells to cisplatin through gene regulation, potentially modulating platinum accumulation both in vitro and in vivo. In summary, EZH2 promotes HB progression through epigenetic silencing and noncanonical signaling pathways. These findings support EZH2's contribution to HB pathogenesis, therefore identifying it as a novel therapeutic target.

Indirect Immunofluorescence (IIF) stained Human Epithelial (HEp-2) cells are considered the gold standard for detecting autoimmune diseases. Accurate cell segmentation, though often viewed as an intermediary step to downstream tasks like classification, significantly enhances overall performance when executed with precision. In this study, we conduct a systematic literature review of HEp-2 cell segmentation techniques, identifying 28 key papers utilizing traditional image processing, machine learning classifiers, deep convolutional neural networks (CNNs), and generative adversarial network (GAN) frameworks. Building on these insights, we benchmark 17 CNN models without pretraining and 8 CNN models pretrained on ImageNet using both Frozen Encoder and Tunable Encoder strategies on the I3A dataset. Cross-validation (CV) and Benjamini-Hochberg (BH) significance correction were employed to ensure statistical rigor in model comparisons. Domain-Specific Pretraining (DSPT) experiments demonstrated performance improvements, particularly for underrepresented classes, while Data Augmentation strategies (DA-1 and DA-2) revealed distinct impacts across model categories. GAN-based segmentation experiments using the top-performing CNN architectures as generators within a Pix2Pix framework revealed performance degradation due to data limitations and adversarial training instabilities. Nonetheless, GANs displayed class-specific improvements in visual alignment of segmentation masks. Results were evaluated comprehensively across eight performance metrics, including Dice, IOU, Accuracy, Precision, Sensitivity, Specificity, AU-ROC and AU-PR. This work offers a robust benchmarking of state-of-the-art CNN, GAN, and Transformer-based models for HEp-2 cell segmentation, providing valuable insights for future research directions, including ensemble approaches, dynamic patch sampling, and diffusion models.

An improved understanding of the human lung necessitates advanced systems models informed by an ever-increasing repertoire of molecular omics, cellular imaging, and pathological datasets. To centralize and standardize information across broad lung research efforts, we expanded the LungMAP.net website into a new gateway portal. This portal connects a broad spectrum of research networks, bulk and single-cell multiomics data, and a diverse collection of image data that span mammalian lung development and disease. The data are standardized across species and technologies using harmonized data and metadata models that leverage recent advances, including those from the Human Cell Atlas, diverse ontologies, and the LungMAP CellCards initiative. To cultivate future discoveries, we have aggregated a diverse collection of single-cell atlases for multiple species (human, rhesus, and mouse) to enable consistent queries across technologies, cohorts, age, disease, and drug treatment. These atlases are provided as independent and integrated queryable datasets, with an emphasis on dynamic visualization, figure generation, reanalysis, cell-type curation, and automated reference-based classification of user-provided single-cell genomics datasets (Azimuth). As this resource grows, we intend to increase the breadth of available interactive interfaces, supported data types, data portals and datasets from LungMAP, and external research efforts.

In the field of histopathology, many studies on the classification of whole slide images (WSIs) using artificial intelligence (AI) technology have been reported. We have studied the disease progression assessment of glioma. Adult-type diffuse gliomas, a type of brain tumor, are classified into astrocytoma, oligodendroglioma, and glioblastoma. Astrocytoma and oligodendroglioma are also called low grade glioma (LGG), and glioblastoma is also called glioblastoma multiforme (GBM). LGG patients frequently have isocitrate dehydrogenase (IDH) mutations. Patients with IDH mutations have been reported to have a better prognosis than patients without IDH mutations. Therefore, IDH mutations are an essential indicator for the classification of glioma. That is why we focused on the IDH1 mutation. In this paper, we aimed to classify the presence or absence of the IDH1 mutation using WSIs and clinical data of glioma patients. Ensemble learning between the WSIs model and the clinical data model is used to classify the presence or absence of IDH1 mutation. By using slide level labels, we combined patch-based imaging information from hematoxylin and eosin (H & E) stained WSIs, along with clinical data using deep image feature extraction and machine learning classifier for predicting IDH1 gene mutation prediction versus wild-type across cohort of 546 patients. We experimented with different deep learning (DL) models including attention-based multiple instance learning (ABMIL) models on imaging data along with gradient boosting machine (LightGBM) for the clinical variables. Further, we used hyperparameter optimization to find the best overall model in terms of classification accuracy. We obtained the highest area under the curve (AUC) of 0.823 for WSIs, 0.782 for clinical data, and 0.852 for ensemble results using MaxViT and LightGBM combination, respectively. Our experimental results indicate that the overall accuracy of the AI models can be improved by using both clinical data and images.

Importance Mental health (MH) issues in children with cerebral palsy (CP) are poorly understood compared with other pediatric populations. Objective To examine MH diagnosis code assignment among children and young adults with CP and compare with typically developing (TD) and chronic condition (CC) pediatric populations. Design, Setting, and Participants This case-control study used International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes to create a CP case set and CC and TD control sets using electronic health record data of children and young adults from a large tertiary care children's hospital in the midwestern United States between 2010 and 2022. Case-control matching was performed to control for demographic factors. Data were analyzed from June to December 2023. Exposures All MH diagnosis codes were mapped to ICD-10-CM and categorized using Clinical Classifications Software Refined (CCSR). Main Outcomes and Measures The incidence rates of MH CCSR categories were calculated. Descriptive and comparative statistics were used to evaluate the significance and odds associated with factors. Results Data from 216 794 individuals (mean [SD] baseline age, 4.3 [5.1] years; 118 562 [55%] male) were analyzed, including 3544 individuals with CP, 142 160 individuals with CC, and 71 080 TD individuals. The CP cohort spread across Gross Motor Function Classification System (GMFCS) levels I (981 individuals [28%]), II (645 individuals [18%]), III (346 individuals [10%]), IV (502 individuals [14%]), and V (618 individuals [17%]). Rates varied significantly for anxiety (824 individuals with CP [23%]; 25 877 individuals with CC [9%]; 6274 individuals with TD [18%]), attention-deficit/hyperactivity disorder (534 individuals with CP [15%]; 22 426 individuals with CC [9%]; 6311 individuals with TD [16%]); conduct or impulse disorder (504 individuals with CP [14%]; 13 209 individuals with CC [5%]; 3715 individuals with TD [9%]), trauma or stress disorders (343 individuals with CP [10%]; 18 229 individuals with CC [8%]; 5329 individuals with TD [13%]), obsessive-compulsive disorder (251 individuals with CP [7%]; 3795 individuals with CC [1%]; 659 individuals with TD [3%]), depression (108 individuals with CP [3%]; 12 224 individuals with CC [5%]; 4007 individuals with TD [9%]), mood disorders (74 individuals with CP [2%]; 4355 individuals with CC [2%]; 1181 individuals with TD [3%]), and suicidal ideation (72 individuals with CP [2%]; 7422 individuals with CC [5%]; 3513 individuals with TD [5%]). There was significant variation in odds of MH diagnoses by GMFCS level (I-II vs III-V: odds ratio [OR], 1.23; 95% CI, 1.09-1.40; P = .001). Among individuals with CP, males were more likely than females to have diagnosis codes for conduct or impulse disorders (OR, 1.41; 95% CI, 1.16-1.73) and attention-deficit/hyperactivity disorder (OR, 1.41 [95% CI, 1.15-1.73]). Black individuals, compared with White individuals, were more likely to have diagnoses for obsessive-compulsive disorder (OR, 1.57 [95% CI, 1.14-2.16]), other mood disorders (OR, 1.85 [95% CI, 1.01-3.38]), and trauma or stress disorders (OR, 1.94 [95% CI, 1.44-2.63]). Odds for trauma or stress disorders were elevated for individuals who identified as other races compared with White individuals (OR, 2.80 [95% CI, 2.03-3.87]). Conclusions and Relevance In this case-control study of children and young adults with CP and matched comparisons, anxiety and conduct or impulse diagnoses were higher in individuals with CP. The lower diagnosis rates of depression and suicidal ideation may indicate underdiagnosis among individuals with CP. There is likely a need for assessment tools that are more suitable for children with CP.

Background In response to severe kidney injury, the kidney epithelium displays remarkable regenerative capabilities driven by adaptable resident epithelial cells. To date, it has been widely considered that the adult kidney lacks multipotent stem cells; thus, the cellular lineages responsible for repairing proximal tubule damage are incompletely understood. Leucine-rich repeats and immunoglobulin-like domain protein 1-expressing cells (Lrig1(+) cells) have been identified as a long-lived cell in various tissues that can induce epithelial tissue repair. Therefore, we hypothesized that Lrig1(+) cells participate in kidney development and tissue regeneration. Methods We investigated the role of Lrig1(+) cells in kidney injury using mouse models. The localization of Lrig1(+) cells in the kidney was examined throughout mouse development. The function of Lrig1(+) progeny cells in AKI repair was examined in vivo using a tamoxifen-inducible Lrig1-specific Cre recombinase-based lineage tracing in three different kidney injury mouse models. In addition, we conducted single-cell RNA sequencing to characterize the transcriptional signature of Lrig1(+) cells and trace their progeny. Results Lrig1(+) cells were present during kidney development and contributed to formation of the proximal tubule and collecting duct structures in mature mouse kidneys. In three-dimensional culture, single Lrig1(+) cells demonstrated long-lasting propagation and differentiated into the proximal tubule and collecting duct lineages. These Lrig1(+) proximal tubule cells highly expressed progenitor-like and quiescence-related genes, giving rise to a novel cluster of cells with regenerative potential in adult kidneys. Moreover, these long-lived Lrig1(+) cells expanded and repaired damaged proximal tubule in response to three types of AKIs in mice. Conclusions These findings highlight the critical role of Lrig1(+) cells in kidney regeneration.

Histopathological image analysis remains at the forefront of computational pathology presenting numerous challenges and demanding tasks, primarily due to the complex nature of tissue structures and the extensive scale of whole slide images (WSIs). Deep learning models have been widely used in histopathology image analysis, especially convolutional neural network (CNN)-based models for classification. However, CNNs have certain limitations due to their small receptive field. Recent works employed adaptations of the classical transformer architecture to visual data [1] [2]. Models such as Vision Transformer (ViT) and Swin Transformer leverage the powerful multi-head self-attention mechanism and have demonstrated comparable or superior performance to state-of-the-art CNN-based classification models. Despite their successes, these models require huge amounts of training data to effectively learn representations as they lack the inherent inductive biases of CNNs. This work compares Vision Transformers with baseline CNN models using a breast cancer histopathological dataset. Further, we employ a novel knowledge-distillation approach to enhance the learning efficiency of vits, When trained with a limited amount of data, Unlike previous works, we aimed to minimize convolution operations when generating patch embeddings to preserve spatial information before reaching the transformer attention layers, we achieved an accuracy of 87.7% for the ViT-base trained as a student of ResNet50, which represents a 1.2% improvement in accuracy over the standalone ViT-base. [3]

The advent of deep learning (DL) and multimodal spatial transcriptomics (ST) has revolutionized cancer research, offering unprecedented insights into tumor biology. This book chapter explores the integration of DL with ST to advance cancer diagnostics, treatment planning, and precision medicine. DL, a subset of artificial intelligence, employs neural networks to model complex patterns in vast datasets, significantly enhancing diagnostic and treatment applications. In oncology, convolutional neural networks excel in image classification, segmentation, and tumor volume analysis, essential for identifying tumors and optimizing radiotherapy. The chapter also delves into multimodal data analysis, which integrates genomic, proteomic, imaging, and clinical data to offer a holistic understanding of cancer biology. Leveraging diverse data sources, researchers can uncover intricate details of tumor heterogeneity, microenvironment interactions, and treatment responses. Examples include integrating MRI data with genomic profiles for accurate glioma grading and combining proteomic and clinical data to uncover drug resistance mechanisms. DL's integration with multimodal data enables comprehensive and actionable insights for cancer diagnosis and treatment. The synergy between DL models and multimodal data analysis enhances diagnostic accuracy, personalized treatment planning, and prognostic modeling. Notable applications include ST, which maps gene expression patterns within tissue contexts, providing critical insights into tumor heterogeneity and potential therapeutic targets. In summary, the integration of DL and multimodal ST represents a paradigm shift towards more precise and personalized oncology. This chapter elucidates the methodologies and applications of these advanced technologies, highlighting their transformative potential in cancer research and clinical practice.

OBJECTIVE The aims of this study were to 1) assess and quantify white matter (WM) microstructural characteristics derived from diffusion tensor imaging (DTI) in children with cerebral palsy (CP) prior to selective dorsal rhizotomy (SDR), and 2) investigate potential associations between WM diffusion properties and gross motor function and spasticity in children with spastic CP who underwent SDR. METHODS This study is a multisite study based on DT images acquired prior to SDR as well as postoperative outcome data. DTI data collected from two sites were harmonized using the ComBat approach to minimize intersite scanner difference. The DTI abnormalities between children with spastic CP and controls were analyzed and correlated with the severity of impaired mobility based on the Gross Motor Function Classification System (GMFCS). The improvement in gross motor function and spasticity after SDR surgery was assessed utilizing the Gross Motor Function Measure-66 (GMFM-66), the Modified Tardieu Scale (MTS), and the modified Ashworth scale (MAS). Alterations in these outcome measures were quantified in association with DTI abnormalities. RESULTS Significant DTI alterations, including lower fractional anisotropy (FA) in the genu of the corpus callosum (gCC) and higher mean diffusivity (MD) in the gCC and posterior limb of the internal capsule (PLIC), were found in children in the SDR group when compared with the age-matched control group (all p < 0.05). Greater DTI alterations (FA in gCC and MD in gCC and PLIC) were associated with lower mobility levels as determined based on GMFCS level (p < 0.05). The pre- to post-SDR improvement in motor function based on GMFM-66 was statistically significant (p = 0.006 and 0.002 at 6-month and 12-month follow-ups, respectively). The SDR efficacy was also identified as improving spasticity in lower-extremity muscle groups assessed with the MTS and MAS. Partial correlation analysis presented a significant association between pre- to post-SDR MTS alteration and DTI abnormalities. CONCLUSIONS The findings in the present study provided initial quantitative evidence to establish the WM microstructural characteristics in children with spastic CP prior to SDR surgery. The study generated data for the association between baseline DTI characteristics and mobility in children with CP prior to SDR surgery. The study also demonstrated SDR efficacy in improving motor function and spasticity based on the GMFM-66, MTS, and MAS, respectively, in association with DTI data.

Species-specific genes are ubiquitous in evolution, with functions ranging from prey paralysis to survival in subzero temperatures. Because they are typically expressed under limited conditions and lack canonical features, such genes may be vastly under-identified, even in humans. Here, we leverage terabytes of human RNA-Seq data to identify thousands of highly-expressed transcripts that do not correspond to any Gencode-annotated gene. Many may be novel ncRNAs although 80% of them contain ORFs that have the potential of encoding proteins unique to Homo sapiens (orphan genes). We validate our findings with independent strand-specific and single-cell RNA-seq datasets. Hundreds of these novel transcripts overlap with deleterious genomic variants; thousands show significant association with disease-specific patient survival. Most are dynamically regulated and accumulate selectively in particular tissues, cell-types, developmental stages, tumors, COVID-19, sex, and ancestries. As such, these transcripts hold potential as diagnostic biomarkers or therapeutic targets. To empower future discovery, we provide a compendium of these huge RNA-Seq expression data, and RiboSeq data, with associated metadata. Further, we supply the gene models for the novel genes as UCSC Genome Browser tracks. ### Competing Interest Statement The authors have declared no competing interest.

Background: Clinical genetic testing is increasingly being utilized to establish a molecular diagnosis to help manage children with cardiomyopathy and to assess the risk of cardiomyopathy among family members. However, as evidence and guidelines evolve, variant classification can change with the potential to impact counseling and family screening. Objectives: The main purpose of this study was to investigate whether variants in cardiomyopathy genes previously interpreted by clinical genetic testing laboratories would be reclassified under current guidelines for the interpretation of sequence variants. Methods: In 211 children enrolled in the Pediatric Cardiomyopathy Registry, we compared the results of previous clinical genetic testing with the results of research testing in 37 cardiomyopathy genes. Results: The mean time difference between initial testing and reinterpretation was 7 years. Using the 2015 American College of Medical Genetics and Genomics guidelines for the interpretation of sequence variants, we found that 18 % of the tested population had a change in variant classification. Ninety-two percent of the initial classifications were performed before the publication of the guidelines, with 82 % of reclassifications resulting in a variant downgrade. Most of these were changes from the pathogenic or likely pathogenic category to a variant of uncertain significance. Reclassification frequency was similar across types of cardiomyopathy. Conclusion: Our results highlight that a portion of variants get downgraded, and periodic reinterpretation of genetic testing results is necessary for all types of cardiomyopathy -particularly for variant interpretations prior to 2015. Importantly, variant reclassification has potential impact on the clinical management of at -risk patients.

Purpose: To characterize physical therapy (PT) dose for children with cerebral palsy (CP) after multi-level surgery (MLS) and examine variation by ambulatory status and surgical burden. Methods: PT dose (Frequency, Intensity, Time, Type) data were extracted from electronic records of children with CP who received outpatient PT the year after MLS. Results: Seventeen children, mean 9 years, female (n=10), ambulatory (n=10), and high surgical burden (n=12) were included. In the year after surgery, 345 visits occurred. Intensity across visits was above average. Time was greatest for pre-functional activities, gait, and transitions/transfers. Types most often delivered were neuromuscular, musculoskeletal, and education/training. Ambulatory children received significantly more visits, higher intensity, and time in pre-functional activities and gait than non-ambulatory children. No differences in type by ambulatory status and PT dose by surgical burden were found. Conclusion: PT dose varied the first year after MLS indicating the need for guidelines by ambulatory status. Video Abstract: Supplemental Digital Content available at: http://links.lww.com/PPT/A516

運動障害を伴う脳性麻痺の治療には日常的な歩行機能評価が極めて重要となる．しかしながら，歩行機能を評価するためには光学式モーションキャプチャなどの高価な機器と高度な専門知識が必要となるため，日常的に歩行機能評価を行うことは容易ではない．そこで本研究では，特殊な装置や専門知識を必要とせず，一般的なカメラによって撮影された歩行動画から歩行機能を評価するシステムの構築を目指す．ここでは研究の第一段階として，スマートフォンなどのカメラにより撮影された被験者の歩行動画から深層学習を用いて各関節位置の時系列変化を推定し，歩行の異常度を評価する手法を提案する．また，本論文では実際の脳性麻痺患者の歩行動画から作成されたデータセットを用いた評価実験を行い，提案法により得られた異常度と臨床現場で用いられる歩行機能指標との関係性についても検討した．実験の結果，提案法により患者の歩行機能の異常度を推定できる可能性が示唆された．

PURPOSE:Through automated electronic health record (EHR) data extraction and analysis, this project systematically quantified actual care delivery for children with cerebral palsy (CP) and evaluated alignment with current evidence-based recommendations. METHODS:Utilizing EHR data for over 8000 children with CP, we developed an approach to define and quantify receipt of optimal care, and pursued proof-of-concept with two children with unilateral CP, Gross Motor Function Classification System (GMFCS) Level II. Optimal care was codified as a cluster of four components including physical medicine and rehabilitation (PMR) care, spasticity management, physical therapy (PT), and occupational therapy (OT). A Receipt of Care Score (ROCS) quantified the degree of adherence to recommendations and was compared with the Pediatric Outcomes Data Collection Instrument (PODCI) and Pediatric Quality of Life Inventory (PEDS QL). RESULTS:The two children (12 year old female, 13 year old male) had nearly identical PMR and spasticity component scores while PT and OT scores were more divergent. Functional outcomes were higher for the child who had higher adjusted ROCS. CONCLUSIONS:ROCSs demonstrate variation in real-world care delivered over time and differentiate between components of care. ROCSs reflect overall function and quality of life. The ROCS methods developed are novel, robust, and scalable and will be tested in a larger sample.IMPLICATIONS FOR REHABILITATIONOptimal practice, with an emphasis on integrated multidisciplinary care, can be defined and quantified utilizing evidence-based recommendations.Receipt of optimal care for childhood cerebral palsy can be scored using existing electronic health record data.Big Data approaches can contribute to the understanding of current care and inform approaches for improved care.

Supplementary Figures 1-3, Tables 1-4 from Activator Protein-1 Transcription Factors Are Associated with Progression and Recurrence of Prostate Cancer

Supplementary Table 1 from Prolonged Exposure to Reduced Levels of Androgen Accelerates Prostate Cancer Progression in Nkx3.1; Pten Mutant Mice