Ward's Hierarchical Clustering Research Articles

Fire regime dynamics in mainland Spain. Part 1: Drivers of change

Fire regimes are evolving worldwide driven by socioeconomic and environmental changes. Understanding the forces behind fire regime dynamics is essential to achieve effective wildfire management and policymaking. The current research belongs to a series of two manuscripts aimed at describing spatial-temporal dynamics of fire regime and its drivers in mainland Spain. In this work, we identified the main transitional pathways of fire regime zones between past (1974–1994) and current (1995–2015) conditions to identify the influence of the main wildfire drivers (demographic potential, climate trends, forest interfaces and topography). Our methodology combined Principal Component Analysis and Ward's hierarchical clustering to identify and spatialize homogenous regions of fire regime on the basis of the main fire regime features: number of fires, burned area, burnt area from lightning-caused fires, area affected by large fires (>100 ha) and seasonality. This procedure was replicated in current and past conditions to extract the most frequent combinations of fire regime typologies, eventually leading to a set of binary response variables (change versus persistence of fire regime). Finally, Random Forest regression was applied to estimate the explanatory performance of fire regime drivers for each transition path.Our results point to an overall decline in fire activity over most of the Spanish territory. The hinterlands and Mediterranean coast experienced the greatest decrease in fire activity. In contrast, winter activity has progressed in the North-western region. According to Random Forest outputs, demographic potential highlights as the main driver of fire regime change (both regressive and progressive), followed by climate dynamics (temperature and rainfall trends) and topographic features. In turn, Wildland-Agricultural Interface (WAI) and Wildland-Urban Interface (WUI) were also behind several noticeable trajectories as is the case of winter fire progression (WAI) or moderate increase in fire incidence linked to human-caused fires (WUI).

Fire regime dynamics in mainland Spain. Part 2: A near-future prospective of fire activity

The current research belongs to a series of two manuscripts aiming at describing spatial-temporal dynamics of fire regime and its drivers in Spain. In this work, we present the first attempt to produce a spatial-temporal delimitation of homogeneous fire regime zones in Spain providing insights into the near future. The analyses were based on historical fire records; leveraging autoregressive models to project fire features into the near future. We evaluated the spatial extent of homogenous fire regime zones in three different periods: past (1974–1994), current (1995–2015) and future (2016–2036). To do so, we applied Principal Component Analysis and Ward's hierarchical clustering to identify zones of fire regime on the basis of the spatial and temporal arrangement of their main fire features: number of fires, burned area, burnt area from natural-caused fires, incidence of large fires (> 100 ha) and seasonality. Clusters of fire regime were trained in the current period, being later projected into the past and future periods using of k-Nearest Neighbor classification.ARIMA modeling forecasted a shrinkage in all fire features except natural-caused fires that remained stable. Overall, we detected a transition from significant fire incidence in the past towards a situation with moderate impact of fires in the near future. The Mediterranean coast experienced the largest decline in fire activity with few locations maintaining the historical levels of occurrence of large fires. On the other hand, the Northwestern end of Spain depicted a progression towards winter fire activity while still linked to large fires. This pattern persisted in the near future along the northern coast, whereas an intermix of minor fire progression and regression was expected thorough the hinterlands and the Mediterranean.

Variance-based Features for Keyword Extraction in Persian and English Text Documents

This paper address automatic keyword extraction in Persian and English text documents. Generally, for keyword extraction in a text, a weight is assigned to each token and words having higher weights are selected as the keywords. We have proposed four methods for weighting the words and have compared these methods with five previous weighting techniques. The previous methods used in this paper are term frequency (TF), term frequency inverse document frequency (TF-IDF), variance, discriminative feature selection (DFS), and document length normalization based on unit words (LNU). The proposed weighting methods are based on using variance features and include variance to TF-IDF ratio, variance to TF ratio, the intersection of TF and variance, and the intersection of variance and IDF. For evaluation, the documents are clustered using the extracted keywords as feature vectors, and K-means, expectation maximization (EM), and Ward hierarchical clustering methods. The entropy of the clusters and pre-defined classes of the documents are used as the evaluation metric. For the evaluations, we have collected and labelled Persian documents. Results show that our proposed weighting method, variance to TF ratio, has the best performance for Persian. Also, the best entropy is resulted by variance to TD-IDF ratio for English.

Perceived health of patients with common variable immunodeficiency-a cluster analysis.

Common variable immunodeficiency (CVID) is a complex disease with various influences on perceived health, which correlate with different outcomes, including new morbidity and mortality. Our hypothesis was that CVID patients fall into distinct clusters of perceived health which can inform care. Ward hierarchical cluster analysis and K-means cluster analysis were performed on data of 209 CVID patients to identify subgroups regarding their self-reported physical and mental health status, assessed by the physical (PCS) and mental component scores (MCS) of the Short Form-12 (SF-12). Four clusters of CVID-patients were identified. Cluster 1 was the largest cluster, characterized by a relatively high physical and mental health status (44·0%). In contrast, cluster 2 (21·1%) included patients with low physical and mental health status. Clusters 3 and 4 were mixed groups with high mental and low physical health (15·8%) and vice versa (19·1%). Significant differences between the clusters were found for patient-reported outcomes such as work ability and health literacy, but not for CVID-associated complications such as enteropathy, interstitial lung disease, granulomatosis, lymphadenopathy and autoimmune cytopenia or laboratory parameters such as immunoglobulin levels or B cell-based classification. The results suggest different subgroups of CVID patients with contrasting individual needs which, surprisingly, did not differ in clinical or laboratory characteristics. The main finding of this study is that patients with CVID fall into four distinct clusters according to perceived health, which are largely independent of CVID complications.

Mapping improvement options of government services: illustration of ‘concept-mapping’

This paper introduces concept mapping – one form of ‘structured conceptualization’ – through describing and structuring improvement options of Hungarian government service centres (in Hungarian, these service centres are also known as ‘government windows’ or ‘kormanyablakok’). Hungarian service centres are implementations of the ‘one-stop shop’ concept where almost all government service needs (close to 2,500 cases or procedures; the most common ones are about ID cards, passports and vehicle registrations) of citizens can be met independently of their residence. Conse-quently, successful implementation of this concept requires new, service-oriented attitude from the staff and much higher level of technical integration than before. Concept mapping uses inputs from stakeholders (managers and staff members of these centres) to describe the problem domain (service improvement options in this example), using natural language input (Hungarian in this case). These inputs are then grouped (by multi-dimensional scaling and Ward hierarchical clustering) and rated by importance. Inputs from the stakeholders were solicited continually throughout the project to determine the final number of clusters and to develop an action plan for improvement, considering the relative importance of the options. Pattern matching was also used to highlight differences between stake-holder groups.

Vertical Wind Profile Characterization and Identification of Patterns Based on a Shape Clustering Algorithm

Wind power plants are becoming a generally accepted resource in the generation mix of many utilities. At the same time, the size and the power rating of individual wind turbines have increased considerably. Under these circumstances, the sector is increasingly demanding an accurate characterization of vertical wind speed profiles to estimate properly the incoming wind speed at the rotor swept area and, consequently, assess the potential for a wind power plant site. This paper describes a shape-based clustering characterization and visualization of real vertical wind speed data. The proposed solution allows us to identify the most likely vertical wind speed patterns for a specific location based on real wind speed measurements. Moreover, this clustering approach also provides characterization and classification of such vertical wind profiles. This solution is highly suitable for a large amount of data collected by remote sensing equipment, where wind speed values at different heights within the rotor swept area are available for subsequent analysis. The methodology is based on $z$ -normalization, shape-based distance metric solution, and the Ward-hierarchical clustering method. Real vertical wind speed profile data corresponding to a Spanish wind power plant and collected by using commercial Windcube equipment during several months are used to assess the proposed characterization and clustering process, involving more than 100 000 wind speed data values. All analyses have been implemented using open-source R-software. From the results, at least four different vertical wind speed patterns are identified to characterize properly over 90% of the collected wind speed data along the day. Therefore, alternative analytical function criteria should be subsequently proposed for vertical wind speed characterization purposes.

Modified balanced random forest for improving imbalanced data prediction

This paper proposes a Modified Balanced Random Forest (MBRF) algorithm as a classification technique to address imbalanced data. The MBRF process changes the process in a Balanced Random Forest by applying an under-sampling strategy based on clustering techniques for each data bootstrap decision tree in the Random Forest algorithm. To find the optimal performance of our proposed method compared with four clustering techniques, like: K-MEANS, Spectral Clustering, Agglomerative Clustering, and Ward Hierarchical Clustering. The experimental result show the Ward Hierarchical Clustering Technique achieved optimal performance, also the proposed MBRF method yielded better performance compared to the Balanced Random Forest (BRF) and Random Forest (RF) algorithms, with a sensitivity value or true positive rate (TPR) of 93.42%, a specificity or true negative rate (TNR) of 93.60%, and the best AUC accuracy value of 93.51%. Moreover, MBRF also reduced process running time.

National fiscal frameworks in the post-crisis European Union

Research background: The latest economic and financial crisis has seriously injured European Union Member States, affecting the condition of their public finances. In the face of the crisis, the EU made a special effort to increase the effectiveness of national fiscal frameworks, e.g. by improving the compliance with legislation. The post 2009 reforms were aimed at providing a solid economic foundation for the national fiscal frameworks, especially in the high-debt euro area countries.
 Purpose of the article: The goal of this research is twofold. Firstly, it aims to provide an outline of the national fiscal governance in the EU. Secondly, the paper analyzes the changes in the core measures of fiscal governance in the EU between the crisis period and the year 2016 (due to the latest available data) and investigates the similarities in the progress made by the 28 EU countries in restoring balance in public finance.
 Methods: To achieve the goal, the literature review and the analysis of core elements of national fiscal frameworks are provided. In the empirical section the grouping method for all 28 EU countries based on the Ward's agglomerative hierarchical clustering method is employed. The study uses data derived from the AMECO database (in the case of fiscal data) and the European Commission thematic data for quality indexes of particular elements of fiscal governance (numerical fiscal rules, medium-term budgetary frameworks and independent fiscal institutions).
 Findings & Value added: This paper contributes to the literature by, on the one hand, attempting to analyze changes in main fiscal governance measures and, on the other hand, by assessing their link with public finance through employment of the agglomerative clustering method. Based on the results, the conclusion about the importance of the improvement in fiscal frameworks is provided. The analysis shows that countries with better national fiscal framework achieved better results in public finances regardless the macroeconomic conditions.

Lower airway microbiota associates with inflammatory phenotype in severe preschool wheeze

Lower airway microbiota associates with inflammatory phenotype in severe preschool wheeze

Fatty acid analysis in the seeds of 50 Paeonia ostii individuals from the same population

Tree peony seeds are rich in α-linolenic acid (ALA), and the peony seed oil is now being produced in China. Paeonia ostii is the most widely used tree peony species for oil extraction, which is commercially called Fengdan and treated as a single cultivar. Here, 50 P. ostii individuals from the same population in northern China were randomly selected for fatty acids (FAs) analysis. Thirteen FAs were isolated, and the most abundant five were palmitic acid (5.31–6.99%), stearic acid (1.22–2.76%), oleic acid (18.78–28.15%), linoleic acid (11.86–26.10%), and ALA (41.11–57.51%). There were significant individual differences of plants in FA quality and quantity and the linoleic acid content in Plant No. 48 even exceeded the scope of 1–99%. Further statistical analysis indicated that most of the individual FAs, saturated FAs, unsaturated FAs, and total FAs levels showed significant positive correlations to each other, whereas the seed yield per plant was independent and not correlated to the factors mentioned above. Ward's hierarchical clustering results grouped the 50 plants into four clusters based on FA contents and seed yield, and the seven plants in Cluster IV were identified as good candidates for oil production. Our results confirmed that the individual differences did occur in P. ostii and Fengdan cannot be simply treated as one uniform cultivar. Also, these results may help simplify the selection of plants for oil peony breeding and accelerate the development of the oil peony industry.

Parcellating cognitive heterogeneity in early psychosis-spectrum illnesses: A cluster analysis

Cognitive impairment is argued to represent a core feature of psychosis-spectrum illnesses. However, within-diagnosis heterogeneity is common, and risk factors for poor cognition remain to be examined after statistically accounting for heterogeneity. Accordingly, we used a data-driven technique (cluster analysis) to empirically-derive cognitive clusters across diagnoses and examined whether concurrent substance use or a history of a neurodevelopmental/behavioral disorder differed between clusters. Data from 135 young help-seekers (aged 12–30 years) with a psychosis-spectrum illness were retrospectively analyzed. Ward's hierarchical cluster analysis classified three cognitive clusters characterized by: (1) normal-range; (2) mixed; and (3) grossly-impaired performance. Despite mostly comparable clinical and demographic measures, cluster 1 had superior socio-occupational functioning and the highest estimated premorbid IQ, followed sequentially by clusters 2 and 3. Proportions of cannabis and amphetamine users did not differ significantly across clusters, nor did rates of patients with a neurodevelopmental/behavioral disorder history. Cluster 3 was however comprised of fewer ‘risky’ drinkers, possibly reflecting reduced opportunity for social drinking associated with cognitive impairment. Estimated premorbid IQ predicted cluster membership (2 vs. 1 & 3 vs. 1), as did clinician-rated socio-occupational functioning and ‘not being enrolled in school or tertiary education’ (3 vs. 1). Our results suggest that concurrent substance use and history of a neurodevelopmental/behavioral disorder do not adequately explain cluster-level cognitive variance in this sample. Future work should integrate neurobiological measures associated with cognition (e.g. white matter integrity) to discern whether clusters reflect neurobiological subtypes better representative of pathophysiology than present symptom-based classifications.

Asthma Phenotypes Defined From Parameters Obtained During Recovery From a Hospital-Treated Exacerbation

Asthma is a heterogeneous disease with diverse clinical manifestations and inflammatory pathologies that is punctuated by exacerbations. To describe the clinical and inflammatory characteristics of patients with asthma treated in hospital for an acute exacerbation. Data from 320 adult patients receiving treatment for an acute exacerbation of asthma were obtained. In 218 patients with complete data, we used the Ward hierarchical clustering to obtain clusters. Pulmonary function, blood cell counts, sputum cell counts, serum IgE levels, and fractional exhaled nitric oxide were measured on hospital admission. We selected 13 variables with which we performed the Ward minimum-variance hierarchical clustering. Four clusters were defined. Clusters 1 (24.5%) and 3 (36.7%) were characterized by predominantly female patients with asthma with sputum neutrophilia, with cluster 1 associated with a small degree of airflow obstruction and early-onset asthma and cluster 3 with a moderate degree of reduction in FEV1. Clusters 2 (22.0%) and 4 (16.5%) were associated with high sputum eosinophilia and severe airflow obstruction. Cluster 4 was made exclusively of male smoking subjects, whereas cluster 2 was made up of predominantly female nonsmoking subjects with the worst FEV1, forced expiratory flow at 25% to 75% of forced vital capacity (% predicted), and partial pressure of oxygen in arterial blood on admission. There were no differences between clusters in terms of atopy, serum IgE, prevalence of nasal disease, dose of maintenance inhaled corticosteroids, or oral/systemic corticosteroid use and asthma exacerbations. The clusters during recovery from an exacerbation of asthma were distinguished by airflow obstruction and a neutrophilic, eosinophilic, or mixed inflammation. Eosinophilic inflammation was found in smoking and nonsmoking patients with asthma during an exacerbation.

Dental caries clusters among adolescents.

There have been very few longitudinal studies of dental caries in adolescents, and little study of the caries risk factors in this age group. The purpose of this study was to describe different caries trajectories and associated risk factors among members of the Iowa Fluoride Study (IFS) cohort. The IFS recruited a birth cohort from 1992 to 1995, and has gathered dietary, fluoride and behavioural data at least twice yearly since recruitment. Examinations for dental caries were completed when participants were ages 5, 9, 13 and 17years. For this study, only participants with decayed and filled surface (DFS) caries data at ages 9, 13 and 17 were included (N=396). The individual DFS counts at age 13 and the DFS increment from 13 to 17 were used to identify distinct caries trajectories using Ward's hierarchical clustering algorithm. A number of multinomial logistic regression models were developed to predict trajectory membership, using longitudinal dietary, fluoride and demographic/behavioural data from 9 to 17years. Model selection was based on the akaike information criterion (AIC). Several different trajectory schemes were considered, and a three-trajectory scheme-no DFS at age 17 (n=142), low DFS (n=145) and high DFS (n=109)-was chosen to balance sample sizes and interpretability. The model selection process resulted in use of an arithmetic average for dietary variables across the period from 9 to 17years. The multinomial logistic regression model with the best fit included the variables maternal education level, 100% juice consumption, brushing frequency and sex. Other favoured models also included water and milk consumption and home water fluoride concentration. The high caries cluster was most consistently associated with lower maternal education level, lower 100% juice consumption, lower brushing frequency and being female. The use of a clustering algorithm and use of Akaike's Information Criterion (AIC) to determine the best representation of the data were useful means in presenting longitudinal caries data. Findings suggest that high caries incidence in adolescence is associated with lower maternal educational level, less frequent tooth brushing, lower 100% juice consumption and being female.

HETEROGENEITY OF HYPOMETABOLIC BRAIN DYSFUNCTION IN AMNESTIC MCI

Amnestic mild cognitive impairment (aMCI) is a clinical diagnosis with considerable etiologic heterogeneity. While the condition often represents prodromal Alzheimer's disease (AD), patients may also progress to other dementia types, stay stable, or even reverse to normal cognition. Detection of glucose hypometabolism by means of FDG-PET is a very sensitive in vivo marker of neuronal dysfunction, and represents one of the best established methods for early detection of an AD-typical neurodegeneration pattern. However, little is known about the heterogeneity of hypometabolic patterns among subjects with aMCI. We applied Ward's hierarchical clustering to voxel-wise FDG-PET data of 403 aMCI patients from the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort to explore subgroups of distinct brain-wide metabolic patterns in this population. Identified subgroups were compared to 179 cognitively normal (CN) controls and characterized with respect to cognitive characteristics, AV45-PET and CSF-based biomarker abnormalities, and clinical progression over 2 year follow-up. Three aMCI subgroups with markedly differing metabolic profiles were identified (see Figure), including (i) individuals with an “AD-typical” hypometabolic pattern (N=199; 49%), who also showed AD characteristic biomarker abnormalities and highest clinical progression rates, (ii) individuals with a “diffuse cortical” hypometabolism (N=85, 21%), who were characterized by relatively more non-amnestic features, less amyloid pathology, and slower clinical decline, and (iii) patients with largely “normal” metabolism (N=119, 30%), who did not show increased biomarker levels or cognitive decline compared to CN controls. “AD-typical” and “diffuse cortical” subgroups could be further stratified into relatively mild and advanced hypometabolic stages with consistent differences in severity of cognitive impairment and clinical progression rates.

Synoptic conditions leading to extremely warm periods in Western Iran

ABSTRACTThe purpose of this study is the characterization of the synoptic circulation patterns leading to extremely warm periods (EWPs) over the Kurdistan province of Iran. Daily maximum temperature data from 27 meteorological stations were used for the period 1962–2012. Data were interpolated to a 6 × 6 km grid using the geospatial Kriging method. EWPs were defined as a period of at least 6 consecutive days with maximum temperature above the local daily‐based 95th percentile of the 1971–2000 period, with additional criteria for spatial extension. The identified EWPs are distributed through the whole year, with maximum frequencies in winter and summer. Interestingly, the decade 1988–1997 did not record any extreme episode and summer extremes were recorded only before 1983. The associated synoptic patterns were classified into four clusters by applying the Ward's hierarchical clustering method to sea level pressure and Z500 data. Four clusters are identified: two represent warm periods in winter, one those occurring spring, especially in March, and another one mainly reflecting summer heat waves. In general, it is found that EWPs are associated with statistically significant anomalies in regional circulation. A deep trough of geopotential height in 500 hPa over North Africa and a ridge directly over the study area were the more important causes for the occurrence of EWPs. Other synoptic configurations that favour extreme periods in winter and summer include, respectively, the northward expansion of the Sudan low pressure with an associated southerly wind advection and the persistence of anticyclonic circulation at all tropospheric levels enhancing subsidence and adiabatic heating. The spring pattern is the most frequent one (nearly 48% of all cases) and reflects a southeastward extension of the subpolar low pressure that leads to the displacement of the Siberian high to south and southwestern Asia and the accompanied northward expansion of the Sudan low.

Delineation Lithuanian agricultural land for agro-ecological suitability for farming using soil and terrain criteria

Abstract The aim of investigation is to analyse and classify the state of agricultural land affected by naturalbiophysical, that is, soil and terrain-slope, handicaps on its overall agro-ecological suitability for agricultural use. For the classification of land for suitability with respect to sustainable use and efficient protection, the following actual criteria were selected: soil texture, soil drainage and terrainslope. For identifying the relatively homogeneous areas, the Ward hierarchical cluster method was used. According to our estimates, Lithuanian agricultural land with unfavourable soil texture, poorly soil drainage and steep slopes covers an area of 33.59, 4.76 and 1.03% of total agricultural area, respectively. On the basis of functional classification of state of agro-ecological conditions of Lithuania, two orders of suitability (S-suitable, N-not suitable) and five land suitability classes were identified and delineated: S1 (highly suitable) included 10 district municipalities, S2 (moderately suitable) included 12 district municipalities, S3 (marginally suitable) included 15 district municipalities, N1 (currently not suitable) included 10 district municipalities and N2 (permanently not suitable) included 4 district municipalities. S3 occupies the largest (29.80%) share of the Lithuanian territory and N2 the least (5.66%). The land suitable for agriculture means with suitability classes S1, S2 and S3 was found on an area of 2,960,562 ha, which is 81.6% of the total land. In addition, soil texture was the most important factor causing differences in the classes of suitability. On the basis of natural limiting factors from an agronomic and eco-environmental perspective, the optimal sustainable development in agrosphere and a balanced-practical concept of land management and proper land use policy is presented. It is a reasonable base for agroecological regionalisation of Lithuania.

On the Uptake of Flexible Working Arrangements and the Association with Human Resource and Organizational Performance Outcomes

The aim of this study was to identify bundles of flexible working arrangements (FWAs) from data provided by 1,064 organizations in seven EU countries, and to relate bundle membership to demographic variables and human resource (HR) and organizational performance outcomes. Using Ward's hierarchical clustering algorithm we identified four distinct bundles of FWAs based on the uptake of twelve individual FWAs across the sample of organizations. Bundle 1 represents organizations engaging in a high level of annual hours contracts; bundle 2 represents more traditional work practices; bundle 3 represents organizations mainly offering shift‐work and bundle 4 represents organizations with a high uptake of flexi‐time. The demographic profile of organizations recorded across each of the four bundles was significantly different. Finally, significant associations were found between the bundle membership and employee turnover (p < 0.001), absenteeism (p < 0.001) and productivity (p < 0.015). The implications of these results are discussed and directions for future research are proposed.

JCES01.04 Liquid Biopsy in Monitoring Dynamic Changes of Driver Genes in Advanced NSCLC

JCES01.04 Liquid Biopsy in Monitoring Dynamic Changes of Driver Genes in Advanced NSCLC

A New Method of Characterizing Flow Patterns of Vortices and Detecting the Centers of Vortices in a Numerical Wind Field

AbstractA vortex in a wind field is an important aspect of a weather system; vortices often result in hazardous weather, such as rainstorms, windstorms, and typhoons. As the availability of numerical meteorological data increases, traditional manual analysis no longer provides an efficient means of timely analysis of observed and predicted atmospheric vortices. Therefore, a method was proposed to automatically characterize flow patterns of vortices and to detect the centers of vortices in complex wind fields generated from numerical weather prediction (NWP) models. First, a statistical feature was developed to preliminarily filter regional wind data to obtain (anti)cyclonic vortices. Second, flow patterns of ideal axisymmetric wind fields were extracted by analyzing circular data related to wind directions. Third, for actual vortices in a complex wind field, a series of rules and deformation degree indices were constructed to retrieve the provisional centers of vortices. Fourth, the Ward hierarchical clustering algorithm was used to cluster these provisional centers, which were filled up by a dilation operation to cover the core region of the vortex. Finally, the vortices were classified as either cyclones or anticyclones based on their analyzed vorticity, and their global centers were precisely located. Experimental results show that the proposed preprocessing method was more effective than the traditional filtering method and that the features of the flow pattern were stable regardless of the variety in the resolution and scale. It was also proven that the proposed method can be further extended and applied to detecting typhoon centers, for which it was more effective than other currently used methods.

Association of Polysensitization, Allergic Multimorbidity, and Allergy Severity: A Cross-Sectional Study of School Children

Background: Aeroallergen sensitization is related to the coexistence of allergic diseases, but the nature of this relationship is poorly understood. The aim of this study was to clarify the relationship of polysensitization with allergic multimorbidities and the severity of allergic diseases. Methods: This study is a cross-sectional analysis of 3,368 Korean children aged 6-7 years-old. We defined IgE-mediated allergic diseases based on structured questionnaires, and classified the sensitivity to 18 aeroallergens by logistic regression and the Ward hierarchical clustering method. The relationship of polysensitization (positive IgE responses against 2 or more aeroallergens classes) with allergic multimorbidities (coexistence of 2 or more of the following allergic diseases: asthma, rhinitis, eczema, and conjunctivitis) and severity of allergic diseases was determined by ordinal logistic regression analysis. Results: The rate of polysensitization was 13.6% (n = 458, 95% CI 12.4-14.8) and that of allergic multimorbidity was 23.5% (n = 790, 95% CI 22.0-24.9). Children sensitized to more aeroallergens tended to have more allergic diseases (rho = 0.248, p < 0.001), although the agreement between polysensitization and multimorbidity was poor (kappa = 0.11, p < 0.001). The number allergen classes to which a child was sensitized increased the risk of wheezing attacks (1 allergen: adjusted odds ratio [aOR] 2.22, 4 or more allergens: aOR 9.39), absence from school (1 allergen: aOR 1.96, 3 allergens: aOR 2.08), and severity of nasal symptoms (1 allergen: aOR 1.61, 4 or more allergens: aOR 4.38). Conclusion: Polysensitization was weakly related to multimorbidity. However, the number of allergens to which a child is sensitized is related to the severity of IgE-mediated symptoms.