An efficient feature selection method for mobile devices with application to activity recognition

This paper presents a feature selection method for data classification, which combines a model-based variable selection technique and a fast two-stage subset selection algorithm. The relationship between a specified (and complete) set of candidate features and the class label is modelled using a non-linear full regression model which is linear-in-the-parameters. The performance of a sub-model measured by the sum of the squared-errors (SSE) is used to score the informativeness of the subset of features involved in the sub-model. The two-stage subset selection algorithm approaches a solution sub-model with the SSE being locally minimized. The features involved in the solution sub-model are selected as inputs to support vector machines (SVMs) for classification. The memory requirement of this algorithm is independent of the number of training patterns. This property makes this method suitable for applications executed in mobile devices where physical RAM memory is very limited. An application was developed for activity recognition, which implements the proposed feature selection algorithm and an SVM training procedure. Experiments are carried out with the application running on a PDA for human activity recognition using accelerometer data. A comparison with an information gain based feature selection method demonstrates the effectiveness and efficiency of the proposed algorithm.

TIME SERIES ANALYSIS AS INPUT FOR PREDICTIVE MODELING: PREDICTING CARDIAC ARREST IN A PEDIATRIC INTENSIVE CARE UNIT

The first manuscript, entitled "Time-Series Analysis as Input for Clinical Predictive Modeling: Modeling Cardiac Arrest in a Pediatric ICU" lays out the theoretical background for the project. There are several core concepts presented in this paper. First, traditional multivariate models (where each variable is represented by only one value) provide single point-in-time snapshots of patient status: they are incapable of characterizing deterioration. Since deterioration is consistently identified as a precursor to cardiac arrests, we maintain that the traditional multivariate paradigm is insufficient for predicting arrests. We identify time series analysis as a method capable of characterizing deterioration in an objective, mathematical fashion, and describe how to build a general foundation for predictive modeling using time series analysis results as latent variables. Building a solid foundation for any given modeling task involves addressing a number of issues during the design phase. These include selecting the proper candidate features on which to base the model, and selecting the most appropriate tool to measure them. We also identified several unique design issues that are introduced when time series data elements are added to the set of candidate features. One such issue is in defining the duration and resolution of time series elements required to sufficiently characterize the time series phenomena being considered as candidate features for the predictive model. Once the duration and resolution are established, there must also be explicit mathematical or statistical operations that produce the time series analysis result to be used as a latent candidate feature. In synthesizing the comprehensive framework for building a predictive model based on time series data elements, we identified at least four classes of data that can be used in the model design. The first two classes are shared with traditional multivariate models: multivariate data and clinical latent features. Multivariate data is represented by the standard one value per variable paradigm and is widely employed in a host of clinical models and tools. These are often represented by a number present in a given cell of a table. Clinical latent features derived, rather than directly measured, data elements that more accurately represent a particular clinical phenomenon than any of the directly measured data elements in isolation. The second two classes are unique to the time series data elements. The first of these is the raw data elements. These are represented by multiple values per variable, and constitute the measured observations that are typically available to end users when they review time series data. These are often represented as dots on a graph. The final class of data results from performing time series analysis. This class of data represents the fundamental concept on which our hypothesis is based. The specific statistical or mathematical operations are up to the modeler to determine, but we generally recommend that a variety of analyses be performed in order to maximize the likelihood that a representation of the time series data elements is produced that is able to distinguish between two or more classes of outcomes. The second manuscript, entitled "Building Clinical Prediction Models Using Time Series Data: Modeling Cardiac Arrest in a Pediatric ICU" provides a detailed description, start to finish, of the methods required to prepare the data, build, and validate a predictive model that uses the time series data elements determined in the first paper. One of the fundamental tenets of the second paper is that manual implementations of time series based models are unfeasible due to the relatively large number of data elements and the complexity of preprocessing that must occur before data can be presented to the model. Each of the seventeen steps is analyzed from the perspective of how it may be automated, when necessary. We identify the general objectives and available strategies of each of the steps, and we present our rationale for choosing a specific strategy for each step in the case of predicting cardiac arrest in a pediatric intensive care unit. Another issue brought to light by the second paper is that the individual steps required to use time series data for predictive modeling are more numerous and more complex than those used for modeling with traditional multivariate data. Even after complexities attributable to the design phase (addressed in our first paper) have been accounted for, the management and manipulation of the time series elements (the preprocessing steps in particular) are issues that are not present in a traditional multivariate modeling paradigm. In our methods, we present the issues that arise from the time series data elements: defining a reference time; imputing and reducing time series data in order to conform to a predefined structure that was specified during the design phase; and normalizing variable families rather than individual variable instances. The final manuscript, entitled: "Using Time-Series Analysis to Predict Cardiac Arrest in a Pediatric Intensive Care Unit" presents the results that were obtained by applying the theoretical construct and its associated methods (detailed in the first two papers) to the case of cardiac arrest prediction in a pediatric intensive care unit. Our results showed that utilizing the trend analysis from the time series data elements reduced the number of classification errors by 73%. The area under the Receiver Operating Characteristic curve increased from a baseline of 87% to 98% by including the trend analysis. In addition to the performance measures, we were also able to demonstrate that adding raw time series data elements without their associated trend analyses improved classification accuracy as compared to the baseline multivariate model, but diminished classification accuracy as compared to when just the trend analysis features were added (ie, without adding the raw time series data elements). We believe this phenomenon was largely attributable to overfitting, which is known to increase as the ratio of candidate features to class examples rises. Furthermore, although we employed several feature reduction strategies to counteract the overfitting problem, they failed to improve the performance beyond that which was achieved by exclusion of the raw time series elements. Finally, our data demonstrated that pulse oximetry and systolic blood pressure readings tend to start diminishing about 10-20 minutes before an arrest, whereas heart rates tend to diminish rapidly less than 5 minutes before an arrest.

Mach Bands: multi-scale spatial filtering and co-operative coding of edges and bars

Perception of Mach bands may be explained by spatial filtering ('lateral inhibition') that can be approximated by 2nd derivative computation, and several alternative models have been proposed. To distinguish between them, we used a novel set of ‘generalised Gaussian’ images, in which the sharp ramp-plateau junction of the Mach ramp was replaced by smoother transitions. The images ranged from a slightly blurred Mach ramp to a Gaussian edge and beyond, and also included a sine-wave edge. The probability of seeing Mach Bands increased with the (relative) sharpness of the junction, but was largely independent of absolute spatial scale. These data did not fit the predictions of MIRAGE, nor 2nd derivative computation at a single fine scale. In experiment 2, observers used a cursor to mark features on the same set of images. Data on perceived position of Mach bands did not support the local energy model. Perceived width of Mach bands was poorly explained by a single-scale edge detection model, despite its previous success with Mach edges (Wallis & Georgeson, 2009, Vision Research, 49, 1886-1893). A more successful model used separate (odd and even) scale-space filtering for edges and bars, local peak detection to find candidate features, and the MAX operator to compare odd- and even-filter response maps (Georgeson, VSS 2006, Journal of Vision 6(6), 191a). Mach bands are seen when there is a local peak in the even-filter (bar) response map, AND that peak value exceeds corresponding responses in the odd-filter (edge) maps.

Effects of testosterone undecanoate as a male contraceptive candidate on rat immunological features

Testosterone undecanoate (TU) is under phase III clinical trial as a hormonal male contraceptive in China. Sex hormones can modulate the immune system. Female hormonal contraceptives may affect SIV/HIV-1 transmission. To evaluate the safety of TU and to understand whether long-term use of TU for a male contraceptive affects users' immunological features, adult male rats were treated for a 32-week TU-treated phase at the dose of 20 mg TU/kg body weight and a 24-week recovery phase. The reproductive and immunological parameters of 4-6 rats in each subgroup were examined at the stated time point. The mean sperm count and viability in the treated rats were significantly suppressed (p < 0.01). In the TU-treated group: the mean blood leukocyte and lymphocyte counts; the proliferation indexes of T cells from peripheral blood mononuclear cells (PBMC) and spleen; and, of B cells from spleen, as well as the mean counts of blood T, NK, and B cells decreased in comparison with those of control group. These decreases were not significant (p > 0.01). Similarly, the mean serum IgM, IgG, and IgA levels and complement activity in TU-treated rats were lower than those in control group (p > 0.01), and the changes in the antibody levels of the examined genital secretions were not significant (p > 0.01). The changes in the thickness of urethra epithelium, and in secretory component (SC) expression in genitals were not observed in the treated group. These results demonstrated that long-term supraphysiological TU injection did not obviously affect the examined rat immunological parameters.

Genetic and gene expression analyses of the polycystic ovary syndrome candidate gene fibrillin-3 and other fibrillin family members in human ovaries

Several studies have demonstrated an association between polycystic ovary syndrome (PCOS) and the dinucleotide repeat microsatellite marker D19S884, which is located in intron 55 of the fibrillin-3 (FBN3) gene. Fibrillins, including FBN1 and 2, interact with latent transforming growth factor (TGF)-β-binding proteins (LTBP) and thereby control the bioactivity of TGFβs. TGFβs stimulate fibroblast replication and collagen production. The PCOS ovarian phenotype includes increased stromal collagen and expansion of the ovarian cortex, features feasibly influenced by abnormal fibrillin expression. To examine a possible role of fibrillins in PCOS, particularly FBN3, we undertook tagging and functional single nucleotide polymorphism (SNP) analysis (32 SNPs including 10 that generate non-synonymous amino acid changes) using DNA from 173 PCOS patients and 194 controls. No SNP showed a significant association with PCOS and alleles of most SNPs showed almost identical population frequencies between PCOS and control subjects. No significant differences were observed for microsatellite D19S884. In human PCO stroma/cortex (n = 4) and non-PCO ovarian stroma (n = 9), follicles (n = 3) and corpora lutea (n = 3) and in human ovarian cancer cell lines (KGN, SKOV-3, OVCAR-3, OVCAR-5), FBN1 mRNA levels were approximately 100 times greater than FBN2 and 200–1000-fold greater than FBN3. Expression of LTBP-1 mRNA was 3-fold greater than LTBP-2. We conclude that FBN3 appears to have little involvement in PCOS but cannot rule out that other markers in the region of chromosome 19p13.2 are associated with PCOS or that FBN3 expression occurs in other organs and that this may be influencing the PCOS phenotype.

Discovery of candidate serum proteomic and metabolomic biomarkers in ankylosing spondylitis

Ankylosing Spondylitis (AS) is a common inflammatory rheumatic disease with a predilection for the axial skeleton, affecting 0.2% of the population. Current diagnostic criteria rely on a composite of clinical and radiological changes, with a mean time to diagnosis of 5 to 10 years. In this study we employed nano liquid-chromatography mass spectrometry analysis to detect and quantify proteins and small compounds including endogenous peptides and metabolites in serum from 18 AS patients and nine healthy individuals. We identified a total of 316 proteins in serum, of which 22 showed significant up- or down-regulation (p < 0.05) in AS patients. Receiver operating characteristic analysis of combined levels of serum amyloid P component and inter-α-trypsin inhibitor heavy chain 1 revealed high diagnostic value for Ankylosing Spondylitis (area under the curve = 0.98). We also depleted individual sera of proteins to analyze endogenous peptides and metabolic compounds. We detected more than 7000 molecular features in patients and healthy individuals. Quantitative MS analysis revealed compound profiles that correlate with the clinical assessment of disease activity. One molecular feature identified as a Vitamin D3 metabolite-(23S,25R)-25-hydroxyvitamin D3 26,23-peroxylactone-was down-regulated in AS. The ratio of this vitamin D metabolite versus vitamin D binding protein serum levels was also altered in AS as compared with controls. These changes may contribute to pathological skeletal changes in AS. Our study is the first example of an integration of proteomic and metabolomic techniques to find new biomarker candidates for the diagnosis of Ankylosing Spondylitis

Compact features for birdcall retrieval from environmental acoustic recordings

Bioacoustic data can be used for monitoring animal species diversity. The deployment of acoustic sensors enables acoustic monitoring at large temporal and spatial scales. We describe a content-based birdcall retrieval algorithm for the exploration of large data bases of acoustic recordings. In the algorithm, an event-based searching scheme and compact features are developed. In detail, ridge events are detected from audio files using event detection on spectral ridges. Then event alignment is used to search through audio files to locate candidate instances. A similarity measure is then applied to dimension-reduced spectral ridge feature vectors. The event-based searching method processes a smaller list of instances for faster retrieval. The experimental results demonstrate that our features achieve better success rate than existing methods and the feature dimension is greatly reduced.

Prediction of Candidate Primary Immunodeficiency Disease Genes Using a Support Vector Machine Learning Approach

Screening and early identification of primary immunodeficiency disease (PID) genes is a major challenge for physicians. Many resources have catalogued molecular alterations in known PID genes along with their associated clinical and immunological phenotypes. However, these resources do not assist in identifying candidate PID genes. We have recently developed a platform designated Resource of Asian PDIs, which hosts information pertaining to molecular alterations, protein-protein interaction networks, mouse studies and microarray gene expression profiling of all known PID genes. Using this resource as a discovery tool, we describe the development of an algorithm for prediction of candidate PID genes. Using a support vector machine learning approach, we have predicted 1442 candidate PID genes using 69 binary features of 148 known PID genes and 3162 non-PID genes as a training data set. The power of this approach is illustrated by the fact that six of the predicted genes have recently been experimentally confirmed to be PID genes. The remaining genes in this predicted data set represent attractive candidates for testing in patients where the etiology cannot be ascribed to any of the known PID genes.

Complex trait of acute pancreatitis : Studies of candidate genes and adipokines

Acute pancreatitis (AP) is a common disease. Mild disease resolves spontaneously in a few days. Severe forms of the disease can lead to local complications, necrosis, and abscesses in and around the pancreas. Systemic inflammation in severe AP is associated with distant organ failures. The aim of this study is to identify genetically determined prognostic factors involved in the clinical features of AP. The study employs a candidate-gene approach, and the genes are involved in trysinogen activation in the initiation phase of the disease, as well as in the systemic inflammation as the disease proceeds. The last study examines adipokines, fat-derived hormones characterized with the capacity to modify inflammation. SPINK 1 is a gene coding trypsin activation inhibitor. Mutations N34S and P55N were determined by minisequencing methods in 371 AP patients and in 459 controls. The mutation N34S was more common in AP patients (7.8%) than in controls (2.6%). This suggests that SPINK 1 gene mutation N34S is a risk factor for AP. In the fourth study, in 12 matched pairs of patients with severe and mild AP, levels of adipokines, adiponectin, and leptin were evaluated. Plasma adipokine levels did not differ between patients with mild and severe AP. The results suggest that in AP, adipokine plasma levels are not factors predisposing to organ failures. This study identified the SPINK 1 mutation N34S to be a risk factor for AP in the general population. As AP is a multifactorial disease, and extensive genetic heterogeneity is likely, further identification of genetic factors in the disease requires larger future studies with more advanced genetic study models. Further identification of the patient characteristics associated with organ failures offers another direction of the study to achieve more detailed understanding of the severe form of AP.

Novel 5-aryloxypyrimidine SEN1576 as a candidate for the treatment of Alzheimer's disease

Prefibrillar assembly of amyloid-ß (Aß) is a major event underlying the development of neuropathology and dementia in Alzheimer's disease (AD). This study determined the neuroprotective properties of an orally bioavailable Aß synaptotoxicity inhibitor, SEN1576. Binding of SEN1576 to monomeric Aß 1–42 was measured using surface plasmon resonance. Thioflavin-T and MTT assays determined the ability of SEN1576 to block Aß 1–42-induced aggregation and reduction in cell viability, respectively. In vivo long-term potentiation (LTP) determined effects on synaptic toxicity induced by intracerebroventricular (i.c.v.) injection of cell-derived Aß oligomers. An operant behavioural schedule measured effects of oral administration following i.c.v. injection of Aß oligomers in normal rats. SEN1576 bound to monomeric Aß 1–42, protected neuronal cells exposed to Aß 1–42, reduced deficits in in vivo LTP and behaviour. SEN1576 exhibits the necessary features of a drug candidate for further development as a disease modifying treatment for the early stages of AD-like dementia.

Ankylosing spondylitis: genomic and functional characterization of candidate genes and their repercussion in clinical practice

RESUMO: Introdução: A espondilite anquilosante (EA) é uma doença inflamatória crónica caracterizada pela inflamação das articulações sacroilíacas e da coluna. A anquilose progressiva motiva uma deterioração gradual da função física e da qualidade de vida. O diagnóstico e o tratamento precoces podem contribuir para um melhor prognóstico. Neste contexto, a identificação de biomarcadores, assume-se como sendo muito útil para a prática clínica e representa hoje um grande desafio para a comunidade científica. Objetivos: Este estudo teve como objetivos: 1 - caracterizar a EA em Portugal; 2 - investigar possíveis associações entre genes, MHC e não-MHC, com a suscetibilidade e as características fenotípicas da EA; 3 - identificar genes candidatos associados a EA através da tecnologia de microarray. Material e Métodos: Foram recrutados doentes com EA, de acordo com os critérios modificados de Nova Iorque, nas consultas de Reumatologia dos diferentes hospitais participantes. Colecionaram-se dados demográficos, clínicos e radiológicos e colhidas amostras de sangue periférico. Selecionaram-se de forma aleatória, doentes HLA-B27 positivos, os quais foram tipados em termos de HLA classe I e II por PCR-rSSOP. Os haplótipos HLA estendidos foram estimados pelo algoritmo Expectation Maximization com recurso ao software Arlequin v3.11. As variantes alélicas dos genes IL23R, ERAP1 e ANKH foram estudadas através de ensaios de discriminação alélica TaqMan. A análise de associação foi realizada utilizando testes da Cochrane-Armitage e de regressão linear, tal como implementado pelo PLINK, para variáveis qualitativas e quantitativas, respetivamente. O estudo de expressão génica foi realizado por Illumina HT-12 Whole-Genome Expression BeadChips. Os genes candidatos foram validados usando qPCR-based TaqMan Low Density Arrays (TLDAs). Resultados: Foram incluídos 369 doentes (62,3% do sexo masculino, com idade média de 45,4 ± 13,2 anos, duração média da doença de 11,4 ± 10,5 anos). No momento da avaliação, 49,9% tinham doença axial, 2,4% periférica, 40,9% mista e 7,1% entesopática. A uveíte anterior aguda (33,6%) foi a manifestação extra-articular mais comum. Foram positivos para o HLA-B27, 80,3% dos doentes. Os haplótipo A*02/B*27/Cw*02/DRB1*01/DQB1*05 parece conferir suscetibilidade para a EA, e o A*02/B*27/Cw*01/DRB1*08/DQB1*04 parece conferir proteção em termos de atividade, repercussão funcional e radiológica da doença. Três variantes (2 para IL23R e 1 para ERAP1) mostraram significativa associação com a doença, confirmando a associação destes genes com a EA na população Portuguesa. O mesmo não se verificou com as variantes estudadas do ANKH. Não se verificou associação entre as variantes génicas não-MHC e as manifestações clínicas da EA. Foi identificado um perfil de expressão génica para a EA, tendo sido validados catorze genes - alguns têm um papel bem documentado em termos de inflamação, outros no metabolismo da cartilagem e do osso. Conclusões: Foi estabelecido um perfil demográfico e clínico dos doentes com EA em Portugal. A identificação de variantes génicas e de um perfil de expressão contribuem para uma melhor compreensão da sua fisiopatologia e podem ser úteis para estabelecer modelos com relevância em termos de diagnóstico, prognóstico e orientação terapêutica dos doentes. -----------ABSTRACT: Background: Ankylosing Spondylitis (AS) is a chronic inflammatory disorder characterized by inflammation in the spine and sacroiliac joints leading to progressive joint ankylosis and in progressive deterioration of physical function and quality of life. An early diagnosis and early therapy may contribute to a better prognosis. The identification of biomarkers would be helpful and represents a great challenge for the scientific community. Objectives: The present study had the following aims: 1- to characterize the pattern of AS in Portuguese patients; 2- to investigate MHC and non-MHC gene associations with susceptibility and phenotypic features of AS and; 3- to identify candidate genes associated with AS by means of whole-genome microarray. Material and Methods: AS was defined in accordance to the modified New York criteria and AS cases were recruited from hospital outcares patient clinics. Demographic and clinical data were recorded and blood samples collected. A random group of HLA-B27 positive patients and controls were selected and typed for HLA class I and II by PCR-rSSOP. The extended HLA haplotypes were estimated by Expectation Maximization Algorithm using Arlequin v3.11 software. Genotyping of IL23R, ERAP1 and ANKH allelic variants was carried out with TaqMan allelic discrimination assays. Association analysis was performed using the Cochrane-Armitage and linear regression tests as implemented in PLINK, for dichotomous and quantitative variables, respectively. Gene expression profile was carried out using Illumina HT-12 Whole-Genome Expression BeadChips and candidate genes were validated using qPCR-based TaqMan Low Density Arrays (TLDAs). Results: A total of 369 patients (62.3% male; mean age 45.4±13.2 years; mean disease duration 11.4±10.5 years), were included. Regarding clinical disease pattern, at the time of assessment, 49.9% had axial disease, 2.4% peripheral disease, 40.9% mixed disease and 7.1% isolated enthesopathic disease. Acute anterior uveitis (33.6%) was the most common extra-articular manifestation. 80.3% of AS patients were HLA-B27 positive. The haplotype A*02/B*27/Cw*02/DRB1*01/DQB1*05 seems to confer susceptibility to AS, whereas A*02/B*27/Cw*01/DRB1*08/DQB1*04 seems to provide protection in terms of disease activity, functional and radiological repercussion. Three markers (two for IL23R and one for ERAP1) showed significant single-locus disease associations. Association of these genes with AS in the Portuguese population was confirmed, whereas ANKH markers studied did not show an association with AS. No association was seen between non-MHC genes and clinical manifestations of AS. A gene expression signature for AS was established; among the fourteen validated genes, a number of them have a well-documented inflammatory role or in modulation of cartilage and bone metabolism. Conclusions: A demographic and clinical profile of patients with AS in Portugal was established. Identification of genetic variants of target genes as well as gene expression signatures could provide a better understanding of AS pathophysiology and could be useful to establish models with relevance in terms of susceptibility, prognosis, and potential therapeutic guidance.

Evolving features-algorithms knowledge map to support NIDS data intelligence and learning loop architecting – a generalised approach to NIDS pattern feature space knowledge processing

This paper describes a proposed new approach to the Computer Network Security Intrusion Detection Systems (NIDS) application domain knowledge processing focused on a topic map technology-enabled representation of features of the threat pattern space as well as the knowledge of situated efficacy of alternative candidate algorithms for pattern recognition within the NIDS domain. Thus an integrative knowledge representation framework for virtualisation, data intelligence and learning loop architecting in the NIDS domain is described together with specific aspects of its deployment.

Variation in the autism candidate gene GABRB3 modulates tactile sensitivity in typically developing children

Background: Autism spectrum conditions have a strong genetic component. Atypical sensory sensitivities are one of the core but neglected features of autism spectrum conditions. GABRB3 is a well-characterised candidate gene for autism spectrum conditions. In mice, heterozygous Gabrb3 deletion is associated with increased tactile sensitivity. However, no study has examined if tactile sensitivity is associated with GABRB3 genetic variation in humans. To test this, we conducted two pilot genetic association studies in the general population, analysing two phenotypic measures of tactile sensitivity (a parent-report and a behavioural measure) for association with 43 SNPs in GABRB3. Findings: Across both tactile sensitivity measures, three SNPs (rs11636966, rs8023959 and rs2162241) were nominally associated with both phenotypes, providing a measure of internal validation. Parent-report scores were nominally associated with six SNPs (P <0.05). Behaviourally measured tactile sensitivity was nominally associated with 10 SNPs (three after Bonferroni correction). Conclusions: This is the first human study to show an association between GABRB3 variation and tactile sensitivity. This provides support for the evidence from animal models implicating the role of GABRB3 variation in the atypical sensory sensitivity in autism spectrum conditions. Future research is underway to directly test this association in cases of autism spectrum conditions.

Graph matching : fast candidate elimination using machine learning techniques

Graph matching is an important class of methods in pattern recognition. Typically, a graph representing an unknown pattern is matched with a database of models. If the database of model graphs is large, an additional factor in induced into the overall complexity of the matching process. Various techniques for reducing the influence of this additional factor have been described in the literature. In this paper we propose to extract simple features from a graph and use them to eliminate candidate graphs from the database. The most powerful set of features and a decision tree useful for candidate elimination are found by means of the C4.5 algorithm, which was originally proposed for inductive learning of classication rules. Experimental results are reported demonstrating that effcient candidate elimination can be achieved by the proposed procedure.

A novel microdeletion syndrome at 3q13.31 characterised by developmental delay, postnatal overgrowth, hypoplastic male genitals, and characteristic facial features

Background Congenital deletions affecting 3q11q23 have rarely been reported and only five cases have been molecularly characterised. Genotype. phenotype correlation has been hampered by the variable sizes and breakpoints of the deletions. In this study, 14 novel patients with deletions in 3q11q23 were investigated and compared with 13 previously reported patients. Methods Clinical data were collected from 14 novel patients that had been investigated by high resolution microarray techniques. Molecular investigation and updated clinical information of one cytogenetically previously reported patient were also included. Results The molecular investigation identified deletions in the region 3q12.3q21.3 with different boundaries and variable sizes. The smallest studied deletion was 580 kb, located in 3q13.31. Genotype. phenotype comparison in 24 patients sharing this shortest region of overlapping deletion revealed several common major characteristics including significant developmental delay, muscular hypotonia, a high arched palate, and recognisable facial features including a short philtrum and protruding lips. Abnormal genitalia were found in the majority of males, several having micropenis. Finally, a postnatal growth pattern above the mean was apparent. The 580 kb deleted region includes five RefSeq genes and two of them are strong candidate genes for the developmental delay: DRD3 and ZBTB20. Conclusion A newly recognised 3q13.31 microdeletion syndrome is delineated which is of diagnostic and prognostic value. Furthermore, two genes are suggested to be responsible for the main phenotype.