Identifying acute illness phenotypes via deep temporal interpolation and clustering network on physiologic signatures.

Using clustering analysis for early vital signs, unique patient phenotypes with distinct pathophysiological signatures and clinical outcomes may be revealed and support early clinical decision-making. Phenotyping using early vital signs has proven challenging, as vital signs are typically sampled sporadically. We proposed a novel, deep temporal interpolation and clustering network to simultaneously extract latent representations from irregularly sampled vital signs and derive phenotypes. Four distinct clusters were identified. Phenotype A (18%) had the greatest prevalence of comorbid disease with increased prevalence of prolonged respiratory insufficiency, acute kidney injury, sepsis, and long-term (3-year) mortality. Phenotypes B (33%) and C (31%) had a diffuse pattern of mild organ dysfunction. Phenotype B's favorable short-term clinical outcomes were tempered by the second highest rate of long-term mortality. Phenotype C had favorable clinical outcomes. Phenotype D (17%) exhibited early and persistent hypotension, high incidence of early surgery, and substantial biomarker incidence of inflammation. Despite early and severe illness, phenotype D had the second lowest long-term mortality. After comparing the sequential organ failure assessment scores, the clustering results did not simply provide a recapitulation of previous acuity assessments. This tool may impact triage decisions and have significant implications for clinical decision-support under time constraints and uncertainty.

Two Symptom Clusters Identified in Patients with Acute Heart Failure.

According to this study: Two distinct symptom clusters were identified in patients with acute heart failure: one with severe congestion and the other with mild congestion.Identification of these symptom clusters in the ED could help with symptom management, patient triage, diagnosis, and self-care at home.

Long COVID exhibits clinically distinct phenotypes at 3-6 months post-SARS-CoV-2 infection: results from the P4O2 consortium.

BACKGROUND: Four months after SARS-CoV-2 infection, 22%-50% of COVID-19 patients still experience complaints. Long COVID is a heterogeneous disease and finding subtypes could aid in optimising and developing treatment for the individual patient. METHODS: Data were collected from 95 patients in the P4O2 COVID-19 cohort at 3-6 months after infection. Unsupervised hierarchical clustering was performed on patient characteristics, characteristics from acute SARS-CoV-2 infection, long COVID symptom data, lung function and questionnaires describing the impact and severity of long COVID. To assess robustness, partitioning around medoids was used as alternative clustering. RESULTS: Three distinct clusters of patients with long COVID were revealed. Cluster 1 (44%) represented predominantly female patients (93%) with pre-existing asthma and suffered from a median of four symptom categories, including fatigue and respiratory and neurological symptoms. They showed a milder SARS-CoV-2 infection. Cluster 2 (38%) consisted of predominantly male patients (83%) with cardiovascular disease (CVD) and suffered from a median of three symptom categories, most commonly respiratory and neurological symptoms. This cluster also showed a significantly lower forced expiratory volume within 1 s and diffusion capacity of the lung for carbon monoxide. Cluster 3 (18%) was predominantly male (88%) with pre-existing CVD and diabetes. This cluster showed the mildest long COVID, and suffered from symptoms in a median of one symptom category. CONCLUSIONS: Long COVID patients can be clustered into three distinct phenotypes based on their clinical presentation and easily obtainable information. These clusters show distinction in patient characteristics, lung function, long COVID severity and acute SARS-CoV-2 infection severity. This clustering can help in selecting the most beneficial monitoring and/or treatment strategies for patients suffering from long COVID. Follow-up research is needed to reveal the underlying molecular mechanisms implicated in the different phenotypes and determine the efficacy of treatment.

Joint Bayesian estimation of cell dependence and gene associations in spatially resolved transcriptomic data.

Recent technologies such as spatial transcriptomics, enable the measurement of gene expressions at the single-cell level along with the spatial locations of these cells in the tissue. Spatial clustering of the cells provides valuable insights into the understanding of the functional organization of the tissue. However, most such clustering methods involve some dimension reduction that leads to a loss of the inherent dependency structure among genes at any spatial location in the tissue. This destroys valuable insights of gene co-expression patterns apart from possibly impacting spatial clustering performance. In spatial transcriptomics, the matrix-variate gene expression data, along with spatial coordinates of the single cells, provides information on both gene expression dependencies and cell spatial dependencies through its row and column covariances. In this work, we propose a joint Bayesian approach to simultaneously estimate these gene and spatial cell correlations. These estimates provide data summaries for downstream analyses. We illustrate our method with simulations and analysis of several real spatial transcriptomic datasets. Our work elucidates gene co-expression networks as well as clear spatial clustering patterns of the cells. Furthermore, our analysis reveals that downstream spatial-differential analysis may aid in the discovery of unknown cell types from known marker genes.

Orthogonal multimodality integration and clustering in single-cell data.

Multimodal integration combines information from different sources or modalities to gain a more comprehensive understanding of a phenomenon. The challenges in multi-omics data analysis lie in the complexity, high dimensionality, and heterogeneity of the data, which demands sophisticated computational tools and visualization methods for proper interpretation and visualization of multi-omics data. In this paper, we propose a novel method, termed Orthogonal Multimodality Integration and Clustering (OMIC), for analyzing CITE-seq. Our approach enables researchers to integrate multiple sources of information while accounting for the dependence among them. We demonstrate the effectiveness of our approach using CITE-seq data sets for cell clustering. Our results show that our approach outperforms existing methods in terms of accuracy, computational efficiency, and interpretability. We conclude that our proposed OMIC method provides a powerful tool for multimodal data analysis that greatly improves the feasibility and reliability of integrated data.

Identifying the most representative actigraphy variables reflecting standardized hand function assessments for remote monitoring in children with unilateral cerebral palsy.

BACKGROUND: Accurate assessment of physical activity and motor function in children with cerebral palsy is crucial for determining the effectiveness of interventions. This study aimed to investigate the correlation between real-world activity monitoring outcomes and in-laboratory standardized hand function assessments in children with unilateral cerebral palsy. METHODS: Actigraphy data were collected over 3 days from children aged 4-12 years with unilateral cerebral palsy before in-laboratory assessments. To tackle the high dimensionality and collinearity of actigraphy variables, we first applied hierarchical clustering using the Pearson correlation coefficient as the distance metric and then performed a principal component analysis (PCA) to reduce the dimensionality of our data. RESULTS: Both hierarchical clustering and PCAs revealed a consistent pattern in which magnitude ratio variables (ln[affected side magnitude/less-affected side magnitude]) were more strongly associated with standardized assessments of hand function than with activity time and distance domain variables. Hierarchical clustering analysis identified two distinct clusters of actigraphy variables, with the second cluster primarily consisting of magnitude ratio variables that exhibited the strongest correlation with Melbourne Assessment 2, Pediatric Motor Activity Log, Assisting Hand Assessment, and Manual Ability Classification System level. Principal component 2, primarily representing the magnitude ratio domain, was positively associated with a meaningful portion of subcategories of standardized measures, whereas principal component 1, representing the activity time and distance component, showed limited associations. CONCLUSIONS: The magnitude ratio of actigraphy can provide additional objective information that complements in-laboratory hand function assessment outcomes in future studies of children with unilateral cerebral palsy. TRIAL REGISTRATION IN CLINICALTRIALS.GOV: NCT04904796 (registered prospectively; date of registration: 23/05/2021).

Identifying subgroups in heart failure patients with multimorbidity by clustering and network analysis.

This study presents a workflow for identifying and characterizing patients with Heart Failure (HF) and multimorbidity utilizing data from Electronic Health Records. Multimorbidity, the co-occurrence of two or more chronic conditions, poses a significant challenge on healthcare systems. Nonetheless, understanding of patients with multimorbidity, including the most common disease interactions, risk factors, and treatment responses, remains limited, particularly for complex and heterogeneous conditions like HF. We conducted a clustering analysis of 3745 HF patients using demographics, comorbidities, laboratory values, and drug prescriptions. Our analysis revealed four distinct clusters with significant differences in multimorbidity profiles showing differential prognostic implications regarding unplanned hospital admissions. These findings underscore the considerable disease heterogeneity within HF patients and emphasize the potential for improved characterization of patient subgroups for clinical risk stratification through the use of EHR data.

Clusters of lifestyle behavioral risk factors and their associations with depressive symptoms and stress: evidence from students at a university in Finland.

BACKGROUND: No previous research of university students in Finland assessed lifestyle behavioral risk factors (BRFs), grouped students into clusters, appraised the relationships of the clusters with their mental well-being, whilst controlling for confounders. The current study undertook this task. METHODS: Students at the University of Turku (n = 1177, aged 22.96 ± 5.2 years) completed an online questionnaire that tapped information on sociodemographic variables (age, sex, income sufficiency, accommodation during the semester), four BRFs [problematic alcohol consumption, smoking, food consumption habits, moderate-to-vigorous physical activity (MVPA)], as well as depressive symptoms and stress. Two-step cluster analysis of the BRFs using log-likelihood distance measure categorized students into well-defined clusters. Two regression models appraised the associations between cluster membership and depressive symptoms and stress, controlling for sex, income sufficiency and accommodation during the semester. RESULTS: Slightly more than half the study participants (56.8%) had always/mostly sufficient income and 33% lived with parents/partner. Cluster analysis of BRFs identified three distinct student clusters, namely Cluster 1 (Healthy Group), Cluster 2 (Smokers), and Cluster 3 (Nonsmokers but Problematic Drinkers). Age, sex and MVPA were not different across the clusters, but Clusters 1 and 3 comprised significantly more respondents with always/mostly sufficient income and lived with their parents/partner during the semester. All members in Clusters 1 and 3 were non-smokers, while all Cluster 2 members comprised occasional/daily smokers. Problematic drinking was significantly different between clusters (Cluster 1 = 0%, Cluster 2 = 54%, Cluster 3 = 100%). Cluster 3 exhibited significantly healthier nutrition habits than both other clusters. Regression analysis showed: (1) males and those with sufficient income were significantly less likely to report depressive symptoms or stress; (2) those living with parents/partner were significantly less likely to experience depressive symptoms; (3) compared to Cluster 1, students in the two other clusters were significantly more likely to report higher depressive symptoms; and (4) only students in Cluster 2 were more likely to report higher stress. CONCLUSIONS: BRFs cluster together, however, such clustering is not a clear-cut, all-or-none phenomenon. Students with BRFs consistently exhibited higher levels of depressive symptoms and stress. Educational and motivational interventions should target at-risk individuals including those with insufficient income or living with roommates or alone.

Clustering of HR + /HER2- breast cancer in an Asian cohort is driven by immune phenotypes.

Breast cancer exhibits significant heterogeneity, manifesting in various subtypes that are critical in guiding treatment decisions. This study aimed to investigate the existence of distinct subtypes of breast cancer within the Asian population, by analysing the transcriptomic profiles of 934 breast cancer patients from a Malaysian cohort. Our findings reveal that the HR + /HER2- breast cancer samples display a distinct clustering pattern based on immune phenotypes, rather than conforming to the conventional luminal A-luminal B paradigm previously reported in breast cancers from women of European descent. This suggests that the activation of the immune system may play a more important role in Asian HR + /HER2- breast cancer than has been previously recognized. Analysis of somatic mutations by whole exome sequencing showed that counter-intuitively, the cluster of HR + /HER2- samples exhibiting higher immune scores was associated with lower tumour mutational burden, lower homologous recombination deficiency scores, and fewer copy number aberrations, implicating the involvement of non-canonical tumour immune pathways. Further investigations are warranted to determine the underlying mechanisms of these pathways, with the potential to develop innovative immunotherapeutic approaches tailored to this specific patient population.

Methodological issues of retrospective surveys for measuring mortality of highly clustered diseases: case study of the 2014-16 Ebola outbreak in Bo District, Sierra Leone.

BACKGROUND: There is a lack of empirical data on design effects (DEFF) for mortality rate for highly clustered data such as with Ebola virus disease (EVD), along with a lack of documentation of methodological limitations and operational utility of mortality estimated from cluster-sampled studies when the DEFF is high. OBJECTIVES: The objectives of this paper are to report EVD mortality rate and DEFF estimates, and discuss the methodological limitations of cluster surveys when data are highly clustered such as during an EVD outbreak. METHODS: We analysed the outputs of two independent population-based surveys conducted at the end of the 2014-2016 EVD outbreak in Bo District, Sierra Leone, in urban and rural areas. In each area, 35 clusters of 14 households were selected with probability proportional to population size. We collected information on morbidity, mortality and changes in household composition during the recall period (May 2014 to April 2015). Rates were calculated for all-cause, all-age, under-5 and EVD-specific mortality, respectively, by areas and overall. Crude and adjusted mortality rates were estimated using Poisson regression, accounting for the surveys sample weights and the clustered design. RESULTS: Overall 980 households and 6,522 individuals participated in both surveys. A total of 64 deaths were reported, of which 20 were attributed to EVD. The crude and EVD-specific mortality rates were 0.35/10,000 person-days (95%CI: 0.23-0.52) and 0.12/10,000 person-days (95%CI: 0.05-0.32), respectively. The DEFF for EVD mortality was 5.53, and for non-EVD mortality, it was 1.53. DEFF for EVD-specific mortality was 6.18 in the rural area and 0.58 in the urban area. DEFF for non-EVD-specific mortality was 1.87 in the rural area and 0.44 in the urban area. CONCLUSION: Our findings demonstrate a high degree of clustering; this contributed to imprecise mortality estimates, which have limited utility when assessing the impact of disease. We provide DEFF estimates that can inform future cluster surveys and discuss design improvements to mitigate the limitations of surveys for highly clustered data.

Main findings: For humanitarian organizations it is imperative to document the methodological limitations of cluster surveys and discuss the utility.Added knowledge: This paper adds new knowledge on cluster surveys for highly clustered data such us in Ebola virus disease.Global health impact of policy and action: We provided empirical estimates and discuss design improvements to inform future study.

Analysis and Visualization of Single-Cell Sequencing Data with Scanpy and MetaCell: A Tutorial.

The emergence and development of single-cell RNA sequencing (scRNA-seq) techniques enable researchers to perform large-scale analysis of the transcriptomic profiling at cell-specific resolution. Unsupervised clustering of scRNA-seq data is central for most studies, which is essential to identify novel cell types and their gene expression logics. Although an increasing number of algorithms and tools are available for scRNA-seq analysis, a practical guide for users to navigate the landscape remains underrepresented. This chapter presents an overview of the scRNA-seq data analysis pipeline, quality control, batch effect correction, data standardization, cell clustering and visualization, cluster correlation analysis, and marker gene identification. Taking the two broadly used analysis packages, i.e., Scanpy and MetaCell, as examples, we provide a hands-on guideline and comparison regarding the best practices for the above essential analysis steps and data visualization. Additionally, we compare both packages and algorithms using a scRNA-seq dataset of the ctenophore Mnemiopsis leidyi, which is representative of one of the earliest animal lineages, critical to understanding the origin and evolution of animal novelties. This pipeline can also be helpful for analyses of other taxa, especially prebilaterian animals, where these tools are under development (e.g., placozoan and Porifera).

Assessing treatment effect heterogeneity in the presence of missing effect modifier data in cluster-randomized trials.

Understanding whether and how treatment effects vary across subgroups is crucial to inform clinical practice and recommendations. Accordingly, the assessment of heterogeneous treatment effects based on pre-specified potential effect modifiers has become a common goal in modern randomized trials. However, when one or more potential effect modifiers are missing, complete-case analysis may lead to bias and under-coverage. While statistical methods for handling missing data have been proposed and compared for individually randomized trials with missing effect modifier data, few guidelines exist for the cluster-randomized setting, where intracluster correlations in the effect modifiers, outcomes, or even missingness mechanisms may introduce further threats to accurate assessment of heterogeneous treatment effect. In this article, the performance of several missing data methods are compared through a simulation study of cluster-randomized trials with continuous outcome and missing binary effect modifier data, and further illustrated using real data from the Work, Family, and Health Study. Our results suggest that multilevel multiple imputation and Bayesian multilevel multiple imputation have better performance than other available methods, and that Bayesian multilevel multiple imputation has lower bias and closer to nominal coverage than standard multilevel multiple imputation when there are model specification or compatibility issues.

Pacybara: accurate long-read sequencing for barcoded mutagenized allelic libraries.

MOTIVATION: Long-read sequencing technologies, an attractive solution for many applications, often suffer from higher error rates. Alignment of multiple reads can improve base-calling accuracy, but some applications, e.g. sequencing mutagenized libraries where multiple distinct clones differ by one or few variants, require the use of barcodes or unique molecular identifiers. Unfortunately, sequencing errors can interfere with correct barcode identification, and a given barcode sequence may be linked to multiple independent clones within a given library. RESULTS: Here we focus on the target application of sequencing mutagenized libraries in the context of multiplexed assays of variant effects (MAVEs). MAVEs are increasingly used to create comprehensive genotype-phenotype maps that can aid clinical variant interpretation. Many MAVE methods use long-read sequencing of barcoded mutant libraries for accurate association of barcode with genotype. Existing long-read sequencing pipelines do not account for inaccurate sequencing or nonunique barcodes. Here, we describe Pacybara, which handles these issues by clustering long reads based on the similarities of (error-prone) barcodes while also detecting barcodes that have been associated with multiple genotypes. Pacybara also detects recombinant (chimeric) clones and reduces false positive indel calls. In three example applications, we show that Pacybara identifies and correctly resolves these issues. AVAILABILITY AND IMPLEMENTATION: Pacybara, freely available at https://github.com/rothlab/pacybara, is implemented using R, Python, and bash for Linux. It runs on GNU/Linux HPC clusters via Slurm, PBS, or GridEngine schedulers. A single-machine simplex version is also available.

Probing molecular pathways: Illuminating the connection between COVID-19 and Alzheimer's disease through the endocannabinoid system dynamics.

Our study investigates the molecular link between COVID-19 and Alzheimer's disease (AD). We aim to elucidate the mechanisms by which COVID-19 may influence the onset or progression of AD. Using bioinformatic tools, we analyzed gene expression datasets from the Gene Expression Omnibus (GEO) database, including GSE147507, GSE12685, and GSE26927. Intersection analysis was utilized to identify common differentially expressed genes (CDEGs) and their shared biological pathways. Consensus clustering was conducted to group AD patients based on gene expression, followed by an analysis of the immune microenvironment and variations in shared pathway activities between clusters. Additionally, we identified transcription factor-binding sites shared by CDEGs and genes in the common pathway. The activity of the pathway and the expression levels of the CDEGs were validated using GSE164805 and GSE48350 datasets. Six CDEGs (MAL2, NECAB1, SH3GL2, EPB41L3, MEF2C, and NRGN) were identified, along with a downregulated pathway, the endocannabinoid (ECS) signaling pathway, common to both AD and COVID-19. These CDEGs showed a significant correlation with ECS activity (p < 0.05) and immune functions. The ECS pathway was enriched in healthy individuals' brains and downregulated in AD patients. Validation using GSE164805 and GSE48350 datasets confirmed the differential expression of these genes in COVID-19 and AD tissues. Our findings reveal a potential pathogenetic link between COVID-19 and AD, mediated by CDEGs and the ECS pathway. However, further research and multicenter evidence are needed to translate these findings into clinical applications.

Relationship between the microenvironment and survival in kidney transplantation: a bibliometric analysis from 2013 to 2023.

Background: Kidney transplantation is considered the most effective treatment for end-stage renal failure. Recent studies have shown that the significance of the immune microenvironment after kidney transplantation in determining prognosis of patients. Therefore, this study aimed to conduct a bibliometric analysis to provide an overview of the knowledge structure and research trends regarding the immune microenvironment and survival in kidney transplantation. Methods: Our search included relevant publications from 2013 to 2023 retrieved from the Web of Science core repository and finally included 865 articles. To perform the bibliometric analysis, we utilized tools such as VOSviewer, CiteSpace, and the R package "bibliometrix". The analysis focused on various aspects, including country, author, year, topic, reference, and keyword clustering. Results: Based on the inclusion criteria, a total of 865 articles were found, with a trend of steady increase. China and the United States were the countries with the most publications. Nanjing Medical University was the most productive institution. High-frequency keywords were clustered into 6 areas, including kidney transplantation, transforming growth factor ß, macrophage, antibody-mediated rejection, necrosis factor alpha, and dysfunction. Antibody mediated rejection (2019-2023) was the main area of research in recent years. Conclusion: This groundbreaking bibliometric study comprehensively summarizes the research trends and advances related to the immune microenvironment and survival after kidney transplantation. It identifies recent frontiers of research and highlights promising directions for future studies, potentially offering fresh perspectives to scholars in the field.

Using unsupervised learning to classify inlet water for more stable design of water reuse in industrial parks.

The water reuse facilities of industrial parks face the challenge of managing a growing variety of wastewater sources as their inlet water. Typically, this clustering outcome is designed by engineers with extensive expertise. This paper presents an innovative application of unsupervised learning methods to classify inlet water in Chinese water reuse stations, aiming to reduce reliance on engineer experience. The concept of 'water quality distance' was incorporated into three unsupervised learning clustering algorithms (K-means, DBSCAN, and AGNES), which were validated through six case studies. Of the six cases, three were employed to illustrate the feasibility of the unsupervised learning clustering algorithm. The results indicated that the clustering algorithm exhibited greater stability and excellence compared to both artificial clustering and ChatGPT-based clustering. The remaining three cases were utilized to showcase the reliability of the three clustering algorithms. The findings revealed that the AGNES algorithm demonstrated superior potential application ability. The average purity in six cases of K-means, DBSCAN, and AGNES were 0.947, 0.852, and 0.955, respectively.

MMM and MMMSynth: Clustering of heterogeneous tabular data, and synthetic data generation.

We provide new algorithms for two tasks relating to heterogeneous tabular datasets: clustering, and synthetic data generation. Tabular datasets typically consist of heterogeneous data types (numerical, ordinal, categorical) in columns, but may also have hidden cluster structure in their rows: for example, they may be drawn from heterogeneous (geographical, socioeconomic, methodological) sources, such that the outcome variable they describe (such as the presence of a disease) may depend not only on the other variables but on the cluster context. Moreover, sharing of biomedical data is often hindered by patient confidentiality laws, and there is current interest in algorithms to generate synthetic tabular data from real data, for example via deep learning. We demonstrate a novel EM-based clustering algorithm, MMM ("Madras Mixture Model"), that outperforms standard algorithms in determining clusters in synthetic heterogeneous data, and recovers structure in real data. Based on this, we demonstrate a synthetic tabular data generation algorithm, MMMsynth, that pre-clusters the input data, and generates cluster-wise synthetic data assuming cluster-specific data distributions for the input columns. We benchmark this algorithm by testing the performance of standard ML algorithms when they are trained on synthetic data and tested on real published datasets. Our synthetic data generation algorithm outperforms other literature tabular-data generators, and approaches the performance of training purely with real data.

An effective hotspot mitigation system for Wireless Sensor Networks using hybridized prairie dog with Genetic Algorithm.

Wireless Sensor Networks (WSNs) consist of small, multifunctional nodes distributed across various locations to monitor and record parameters. These nodes store data and transmit signals for further processing, forming a crucial topic of study. Monitoring the network's status in WSN applications using clustering systems is essential. Collaboration among sensors from various domains enhances the precision of localised information reporting. However, nodes closer to the data sink consume more energy, leading to hotspot challenges. To address these challenges, this research employs clustering and optimised routing techniques. The aggregation of information involves creating clusters, further divided into sub-clusters. Each cluster includes a Cluster Head (CH) or Sensor Nodes (SN) without a CH. Clustering inherently optimises CHs' capabilities, enhances network activity, and establishes a systematic network topology. This model accommodates both multi-hop and single-hop systems. This research focuses on selecting CHs using a Genetic Algorithm (GA), considering various factors. While GA possesses strong exploration capabilities, it requires effective management. This research uses Prairie Dog Optimization (PDO) to overcome this challenge. The proposed Hotspot Mitigated Prairie with Genetic Algorithm (HM-PGA) significantly improves WSN performance, particularly in hotspot avoidance. With HM-PGA, it achieves a network lifetime of 20913 milliseconds and 310 joules of remaining energy. Comparative analysis with existing techniques demonstrates the superiority of the proposed approach.

Eating Habits and Lifestyle Factors Related to Childhood Obesity Among Children Aged 5-6 Years: Cluster Analysis of Panel Survey Data in Korea.

BACKGROUND: Childhood obesity has emerged as a major health issue due to the rapid growth in the prevalence of obesity among young children worldwide. Establishing healthy eating habits and lifestyles in early childhood may help children gain appropriate weight and further improve their health outcomes later in life. OBJECTIVE: This study aims to classify clusters of young children according to their eating habits and identify the features of each cluster as they relate to childhood obesity. METHODS: A total of 1280 children were selected from the Panel Study on Korean Children. Data on their eating habits (eating speed, mealtime regularity, consistency of food amount, and balanced eating), sleep hours per day, outdoor activity hours per day, and BMI were obtained. We performed a cluster analysis on the children's eating habits using k-means methods. We conducted ANOVA and chi-square analyses to identify differences in the children's BMI, sleep hours, physical activity, and the characteristics of their parents and family by cluster. RESULTS: At both ages (ages 5 and 6 years), we identified 4 clusters based on the children's eating habits. Cluster 1 was characterized by a fast eating speed (fast eaters); cluster 2 by a slow eating speed (slow eaters); cluster 3 by irregular eating habits (poor eaters); and cluster 4 by a balanced diet, regular mealtimes, and consistent food amounts (healthy eaters). Slow eaters tended to have the lowest BMI (P<.001), and a low proportion had overweight and obesity at the age of 5 years (P=.03) and 1 year later (P=.005). There was a significant difference in sleep time (P=.01) and mother's education level (P=.03) at the age of 5 years. Moreover, there was a significant difference in sleep time (P=.03) and the father's education level (P=.02) at the age of 6 years. CONCLUSIONS: Efforts to establish healthy eating habits in early childhood may contribute to the prevention of obesity in children. Specifically, providing dietary guidance on a child's eating speed can help prevent childhood obesity. This research suggests that lifestyle modification could be a viable target to decrease the risk of childhood obesity and promote the development of healthy children. Additionally, we propose that future studies examine long-term changes in obesity resulting from lifestyle modifications in children from families with low educational levels.

ProDiv: Prototype-driven consistent pseudo-bag division for whole-slide image classification.

BACKGROUND AND OBJECTIVE: Pathology image classification is one of the most essential auxiliary processes in cancer diagnosis. To overcome the problem of inadequate Whole-Slide Image (WSI) samples with weak labels, pseudo-bag-based multiple instance learning (MIL) methods have attracted wide attention in pathology image classification. In this type of method, the division scheme of pseudo-bags is usually a primary factor affecting classification performance. In order to improve the division of WSI pseudo-bags on existing random/clustering approaches, this paper proposes a new Prototype-driven Division (ProDiv) scheme for the pseudo-bag-based MIL classification framework on pathology images. METHODS: This scheme first designs an attention-based method to generate a bag prototype for each slide. On this basis, it further groups WSI patch instances into a series of instance clusters according to the feature similarities between the prototype and patches. Finally, pseudo-bags are obtained by randomly combining the non-overlapping patch instances of different instance clusters. Moreover, the design scheme of our ProDiv considers practicality, and it could be smoothly assembled with almost all the MIL-based WSI classification methods in recent years. RESULTS: Empirical results show that our ProDiv, when integrated with several existing methods, can deliver classification AUC improvements of up to 7.3% and 10.3%, respectively on two public WSI datasets. CONCLUSIONS: ProDiv could almost always bring obvious performance improvements to compared MIL models on typical metrics, which suggests the effectiveness of our scheme. Experimental visualization also visually interprets the correctness of the proposed ProDiv.