Application of data mining techniques in the prediction of coronary artery disease : use of anaesthesia time-series and patient risk factor data

The high morbidity and mortality associated with atherosclerotic coronary vascular disease (CVD) and its complications are being lessened by the increased knowledge of risk factors, effective preventative measures and proven therapeutic interventions. However, significant CVD morbidity remains and sudden cardiac death continues to be a presenting feature for some subsequently diagnosed with CVD. Coronary vascular disease is also the leading cause of anaesthesia related complications. Stress electrocardiography/exercise testing is predictive of 10 year risk of CVD events and the cardiovascular variables used to score this test are monitored peri-operatively. Similar physiological time-series datasets are being subjected to data mining methods for the prediction of medical diagnoses and outcomes. This study aims to find predictors of CVD using anaesthesia time-series data and patient risk factor data. Several pre-processing and predictive data mining methods are applied to this data. Physiological time-series data related to anaesthetic procedures are subjected to pre-processing methods for removal of outliers, calculation of moving averages as well as data summarisation and data abstraction methods. Feature selection methods of both wrapper and filter types are applied to derived physiological time-series variable sets alone and to the same variables combined with risk factor variables. The ability of these methods to identify subsets of highly correlated but non-redundant variables is assessed. The major dataset is derived from the entire anaesthesia population and subsets of this population are considered to be at increased anaesthesia risk based on their need for more intensive monitoring (invasive haemodynamic monitoring and additional ECG leads). Because of the unbalanced class distribution in the data, majority class under-sampling and Kappa statistic together with misclassification rate and area under the ROC curve (AUC) are used for evaluation of models generated using different prediction algorithms. The performance based on models derived from feature reduced datasets reveal the filter method, Cfs subset evaluation, to be most consistently effective although Consistency derived subsets tended to slightly increased accuracy but markedly increased complexity. The use of misclassification rate (MR) for model performance evaluation is influenced by class distribution. This could be eliminated by consideration of the AUC or Kappa statistic as well by evaluation of subsets with under-sampled majority class. The noise and outlier removal pre-processing methods produced models with MR ranging from 10.69 to 12.62 with the lowest value being for data from which both outliers and noise were removed (MR 10.69). For the raw time-series dataset, MR is 12.34. Feature selection results in reduction in MR to 9.8 to 10.16 with time segmented summary data (dataset F) MR being 9.8 and raw time-series summary data (dataset A) being 9.92. However, for all time-series only based datasets, the complexity is high. For most pre-processing methods, Cfs could identify a subset of correlated and non-redundant variables from the time-series alone datasets but models derived from these subsets are of one leaf only. MR values are consistent with class distribution in the subset folds evaluated in the n-cross validation method. For models based on Cfs selected time-series derived and risk factor (RF) variables, the MR ranges from 8.83 to 10.36 with dataset RF_A (raw time-series data and RF) being 8.85 and dataset RF_F (time segmented time-series variables and RF) being 9.09. The models based on counts of outliers and counts of data points outside normal range (Dataset RF_E) and derived variables based on time series transformed using Symbolic Aggregate Approximation (SAX) with associated time-series pattern cluster membership (Dataset RF_ G) perform the least well with MR of 10.25 and 10.36 respectively. For coronary vascular disease prediction, nearest neighbour (NNge) and the support vector machine based method, SMO, have the highest MR of 10.1 and 10.28 while logistic regression (LR) and the decision tree (DT) method, J48, have MR of 8.85 and 9.0 respectively. DT rules are most comprehensible and clinically relevant. The predictive accuracy increase achieved by addition of risk factor variables to time-series variable based models is significant. The addition of time-series derived variables to models based on risk factor variables alone is associated with a trend to improved performance. Data mining of feature reduced, anaesthesia time-series variables together with risk factor variables can produce compact and moderately accurate models able to predict coronary vascular disease. Decision tree analysis of time-series data combined with risk factor variables yields rules which are more accurate than models based on time-series data alone. The limited additional value provided by electrocardiographic variables when compared to use of risk factors alone is similar to recent suggestions that exercise electrocardiography (exECG) under standardised conditions has limited additional diagnostic value over risk factor analysis and symptom pattern. The effect of the pre-processing used in this study had limited effect when time-series variables and risk factor variables are used as model input. In the absence of risk factor input, the use of time-series variables after outlier removal and time series variables based on physiological variable values’ being outside the accepted normal range is associated with some improvement in model performance.

The expression of ADAM-9 and -10 in prostate cancer and their regulation by dihydrotestosterone, insulin-like growth Factor-1 and epidermal growth factor

Prostrate Cancer(PCa)is the most common cause of cancer death amongst Western males. PCa occurs in two distinct stages. In its early stage, growth and development is dependent primarily on male sex hormones (androgens) such as testosterone, although other growth factors have roles maintaining PCa cell survival in this stage. In the later stage of PCa development, growth and.maintenance is independent of androgen stimulation and growth factors including Insulin-like Growth Factor -1 (IGf.:·l) and Epidermal Growth Factor (EGF) are thought to have more crucial roles in cell survival and PCa progression. PCa, in its late stages, is highly aggressive and metastatic, that is, tumorigenic cells migrate from the primary site of the body (prostate) and travel via the systemic and lymphatic circulation, residing and colonising in the bone, lymph node, lung, and in more rare cases, the brain. Metastasis involves both cell migration and tissue degradation activities. The degradation of the extracellular matrix (ECM), the tissue surrounding the organ, is mediated in part by members of a family of 26 proteins called the Matrix Metalloproteases (MMPs), whilst ceil adhesion molecules, of which proteins known as Integrins are included, mediate ce11 migration. A family of proteins known as the ADAMs (A Disintegrin . And Metalloprotease domain) were a recently characterised family at the commencement of this study and now comprise 34 members. Because of their dual nature, possessing an active metaiioprotease domain, homologous to that of the MMPs, and an integrin-binding domain capable of regulating cell-cell and cell-ECM contacts, it was thought likely that members of the ADAMs family may have implications for the progression of aggressive cancers such as those ofthe prostate. This study focussed on two particular ADAMs -9 and -10. ADAM-9 has an active metalloprotease domain, which has been shown to degrade constituents of the ECM, including fibronectin, in vitro. It also has an integrin-binding capacity through association with key integrins involved in PCa progression, such as a6~1. ADAM-10 has no such integrin binding activities, but its bovine orthologue, MADM, is able to degrade coHagen type IV, a major component of basement membranes. It is likely human ADAM-10 has the same activity. It is also known to cleave Ll -a protein involved in cell anchorage activities - and collagen type XVII - which is a principal component of the hemidesmosomes of cellular tight junctions. The cleavage of these proteins enables the cell to be released from the surrounding environment and commence migratory activities, as required in metastasis. Previous studies in this laboratory showed the mRNA expression of the five ADAMs -9,- 10, -11, -15 and -17 in PCa cell lines, characteristic of androgen-dependent and androgen independent disease. These studies were furthered by the characterisation of AD AM-9, -10 and -17 mRNA regulation by Dihydrotestosterone (DHT) in the androgen-responsive cell line (LNCaP). ADAM-9 and -10 mRNA levels were elevated in response to DHT stimulation. Further to these observations, the expression of ADAM-9 and -10 was shown in primary prostate biopsies from patients with PCa. ADAM-1 0 was expressed in the cytoplasm and on the ceH membrane in epithelial and basal cells ofbenign prostate glands, but in high-grade PCa glands, ADAM-I 0 expression was localised to the nucleus and its expression levels appeared to be elevated when compared to low-grade PCa glands. These studies provided a strong background for the hypothesis that ADAM-9 and -10 have key roles in the development ofPCa and provided a basis for further studies.The aims of this study were to: 1) characterise the expression, localisation and levels, of ADAM-9 and -10 mRNA and protein in cell models representing characteristics of normal through androgen-dependent to androgen-independent PCa, as well as to expand the primary PCa biopsy data for ADAM-9 and ADAM-10 to encompass PCa bone metastases 2) establish an in vitro cell system, which could express elevated levels of ADAM-1 0 so that functional cell-based assays such as cell migration, invasion and attachment could be carried out, and 3) to extend the previous hormonal regulation data, to fully characterise the response of ADAM-9 and -10 mRNA and protein levels to DHT, IGF-1, DHT plus IGF-1 and EGF in the hormonal/growth factor responsive cell line LNCaP. For aim 1 (expression of ADAM-9 and -10 mRNA and protein), ADAM-9 and -10 mRNA were characterised by R T -PCR, while their protein products were analysed by Western blot. Both ADAM-9 and -10 mRNA and protein were expressed at readily detectable levels across progressively metastatic PCa cell lines model that represent characteristics of low-grade,. androgen-dependent (LNCaP and C4) to high-grade, androgen-independent (C4-2 and C4-2B) PCa. When the non-tumorigenic prostate cell line RWPE-1 was compared with the metastatic PCa cell line PC-3, differential expression patterns were seen by Western blot analysis. For ADAM-9, the active form was expressed at higher levels in RWPE-1, whilst subcellular fractionation showed that the active form of ADAM-9 was predominantly located in the cell nucleus. For ADAM-I 0, in both of the cell Jines, a nuclear specific isoform of the mature, catalytically active ADAM-I 0 was found. This isoforrn differed by -2 kDa in Mr (smaller) than the cytoplasmic specific isoform. Unprocessed ADAM-I 0 was readily detected in R WPE-1 cell lines but only occasionally detected in PC-3 cell lines. Immunocytochemistry using ADAM-9 and -10 specific antibodies confirmed nuclear, cytoplasmic and membrane expression of both ADAMs in these two cell lines. To examine the possibility of ADAM-9 and -10 being shed into the extracellular environment, membrane vesicles that are constitutively shed from the cell surface and contain membrane-associated proteins were collected from the media of the prostate cell lines RWPE-1, LNCaP and PC-3. ADAM-9 was readily detectable in RWPE- 1 and LNCaP cell membrane vesicles by Western blot analysis, but not in PC-3 cells, whilst the expression of ADAM-I 0 was detected in shed vesicles from each of these prostate cell lines. By Laser Capture Microdissection (LCM), secretory epithelial cells of primary prostate gland biopsies were isolated from benign and malignant glands. These secretory cells, by Western blot analysis, expressed similar Mr bands for ADAM-9 and -10 that were found in PCa cell lines in vitro, indicating that the nuclear specific isoforrn of ADAM-I 0 was present in PCa primary tumours and may represent the predominantly nuclear form of ADAM-I 0 expression, previously shown in high-grade PCa by immunohistochemistry (IHC). ADAM-9 and -10 were also examined by IHC in bone metastases taken from PCa patients at biopsy. Both ADAMs could be detected at levels similar to those shown for Prostate Specific Antigen (PSA) in these biopsies. Furthermore, both ADAM-9 and -10 were predominantly membrane- bound with occasional nuclear expression. For aim 2, to establish a cell system that over-expressed levels of ADAM-10, two fulllength ADAM-I 0 mammalian expression vectors were constructed; ADAM-I 0 was cloned into pcDNA3.1, which contains a CMV promoter, and into pMEP4, containing an inducible metallothionine promoter, whose activity is stimulated by the addition of CdC}z. The efficiency of these two constructs was tested by way of transient transfection in the PCa cell line PC-3, whilst the pcDNA3.1 construct was also tested in the RWPE-1 prostate cell line. Resultant Western blot analysis for all transient transfection assays showed that levels of ADAM-I 0 were not significantly elevated in any case, when compared to levels of the housekeeping gene ~-Tubulin, despite testing various levels of vector DNA, and, for pMEP4, the induction of the transfected cell system with different degrees of stimulation with CdCh to activate the metallothionine promoter post-transfection. Another study in this laboratory found similar results when the same full length ADAM-10 sequence was cloned into a Green Fluorescent Protein (GFP) expressing vector, as no fluorescence was observed by means of transient tran sfection in the same, and other, PCa cell lines. It was hypothesised that the Kozak sequence included in the full-length construct (human ADAMI 0 naturally occurring sequence) is not strong enough to initiate translation in an artificial system, in cells, which, as described in Aim 1, are already expressing readily detectable levels of endogenous ADAM-10. As a result, time constraints prevented any further progress with Aim 2 and functional studies including cell attachment, invasion and migration were unable to be explored. For Aim 3, to characterise the response of ADAM-9 and -10 mRNA and protein levels to DHT, IGF-1, DHT plus IGF-1 and EGF in LNCaP cells, the levels of ADAM-9 and -10 mRNA were not stimulated by DHT or IGF-I alone, despite our previous observations that initially characterised ADAM-9 and -10 mRNA as being responsive to DHT. However, IGF-1 in synergy with DHT did significantly elevate mRNA levels ofboth ADAMs. In the case of ADAM-9 and -10 protein, the same trends of stimulation as found at the rnRNA level were shown by Western blot analysis when ADAM-9 and -10 signal intensity was normalised with the housekeeping protein ~-Tubulin. For EGF treatment, both ADAM-9 and -10 mRNA and protein levels were significantly elevated, and further investigation vm found this to be the case for each of these ADAMs proteins in the nuclear fractions of LNCaP cells. These studies are the first to describe extensively, the expression and hormonal/growth factor regulation of two members of the ADAMs family ( -9 and -1 0) in PCa. These observations imply that the expression of ADAM-9 and -10 have varied roles in PCa whilst it develops from androgen-sensitive (early stage disease), through to an androgeninsensitive (late-stage), metastatic disease. Further studies are now required to investigate the several key areas of focus that this research has revealed, including: • Investigation of the cellular mechanisms that are involved in actively transporting the ADAMs to the cell's nuclear compartment and the ADAMs functional roles in the cell nucleus. • The construction of a full-length human ADAM-10 mammalian expression construct with the introduction of a new Kozak sequence, that elevates ADAM-I 0 expression in an in vitro cell system are required, so that functional assays such as cell invasion, migration and attachment may be carried out to fmd the functional consequences of ADAM expression on cellular behaviour. • The regulation studies also need to be extended by confirming the preliminary observations that the nuclear levels of ADAMs may also be elevated by hormones and growth factors such as DHT, IGF-1 and EGF, as well as the regulation of levels of plasma membrany vesicle associated ADAM expression. Given the data presented in this study, it is likely the ADAMs have differential roles throughout the development of PCa due to their differential cellular localisation and synergistic growth-factor regulation. These observations, along with those further studies outlined above, are necessary in identifying these specific components ofPCa metastasis to which the ADAMs may contribute.

Applying SVMs and weight-based factor analysis to unsupervised adaptation for speaker verification

This paper presents an extended study on the implementation of support vector machine(SVM) based speaker verification in systems that employ continuous progressive model adaptation using the weight-based factor analysis model. The weight-based factor analysis model compensates for session variations in unsupervised scenarios by incorporating trial confidence measures in the general statistics used in the inter-session variability modelling process. Employing weight-based factor analysis in Gaussian mixture models (GMM) was recently found to provide significant performance gains to unsupervised classification. Further improvements in performance were found through the integration of SVM-based classification in the system by means of GMM supervectors. This study focuses particularly on the way in which a client is represented in the SVM kernel space using single and multiple target supervectors. Experimental results indicate that training client SVMs using a single target supervector maximises performance while exhibiting a certain robustness to the inclusion of impostor training data in the model. Furthermore, the inclusion of low-scoring target trials in the adaptation process is investigated where they were found to significantly aid performance.

Predicting residue-wise contact orders in proteins by support vector regression

Background The residue-wise contact order (RWCO) describes the sequence separations between the residues of interest and its contacting residues in a protein sequence. It is a new kind of one-dimensional protein structure that represents the extent of long-range contacts and is considered as a generalization of contact order. Together with secondary structure, accessible surface area, the B factor, and contact number, RWCO provides comprehensive and indispensable important information to reconstructing the protein three-dimensional structure from a set of one-dimensional structural properties. Accurately predicting RWCO values could have many important applications in protein three-dimensional structure prediction and protein folding rate prediction, and give deep insights into protein sequence-structure relationships. Results We developed a novel approach to predict residue-wise contact order values in proteins based on support vector regression (SVR), starting from primary amino acid sequences. We explored seven different sequence encoding schemes to examine their effects on the prediction performance, including local sequence in the form of PSI-BLAST profiles, local sequence plus amino acid composition, local sequence plus molecular weight, local sequence plus secondary structure predicted by PSIPRED, local sequence plus molecular weight and amino acid composition, local sequence plus molecular weight and predicted secondary structure, and local sequence plus molecular weight, amino acid composition and predicted secondary structure. When using local sequences with multiple sequence alignments in the form of PSI-BLAST profiles, we could predict the RWCO distribution with a Pearson correlation coefficient (CC) between the predicted and observed RWCO values of 0.55, and root mean square error (RMSE) of 0.82, based on a well-defined dataset with 680 protein sequences. Moreover, by incorporating global features such as molecular weight and amino acid composition we could further improve the prediction performance with the CC to 0.57 and an RMSE of 0.79. In addition, combining the predicted secondary structure by PSIPRED was found to significantly improve the prediction performance and could yield the best prediction accuracy with a CC of 0.60 and RMSE of 0.78, which provided at least comparable performance compared with the other existing methods. Conclusion The SVR method shows a prediction performance competitive with or at least comparable to the previously developed linear regression-based methods for predicting RWCO values. In contrast to support vector classification (SVC), SVR is very good at estimating the raw value profiles of the samples. The successful application of the SVR approach in this study reinforces the fact that support vector regression is a powerful tool in extracting the protein sequence-structure relationship and in estimating the protein structural profiles from amino acid sequences.

i-vector based speaker recognition on short utterances

Robust speaker verification on short utterances remains a key consideration when deploying automatic speaker recognition, as many real world applications often have access to only limited duration speech data. This paper explores how the recent technologies focused around total variability modeling behave when training and testing utterance lengths are reduced. Results are presented which provide a comparison of Joint Factor Analysis (JFA) and i-vector based systems including various compensation techniques; Within-Class Covariance Normalization (WCCN), LDA, Scatter Difference Nuisance Attribute Projection (SDNAP) and Gaussian Probabilistic Linear Discriminant Analysis (GPLDA). Speaker verification performance for utterances with as little as 2 sec of data taken from the NIST Speaker Recognition Evaluations are presented to provide a clearer picture of the current performance characteristics of these techniques in short utterance conditions.

Learned ultra-wideband RADAR sensor model for augmented LIDAR-based traversability mapping in vegetated environments

In vegetated environments, reliable obstacle detection remains a challenge for state-of-the-art methods, which are usually based on geometrical representations of the environment built from LIDAR and/or visual data. In many cases, in practice field robots could safely traverse through vegetation, thereby avoiding costly detours. However, it is often mistakenly interpreted as an obstacle. Classifying vegetation is insufficient since there might be an obstacle hidden behind or within it. Some Ultra-wide band (UWB) radars can penetrate through vegetation to help distinguish actual obstacles from obstacle-free vegetation. However, these sensors provide noisy and low-accuracy data. Therefore, in this work we address the problem of reliable traversability estimation in vegetation by augmenting LIDAR-based traversability mapping with UWB radar data. A sensor model is learned from experimental data using a support vector machine to convert the radar data into occupancy probabilities. These are then fused with LIDAR-based traversability data. The resulting augmented traversability maps capture the fine resolution of LIDAR-based maps but clear safely traversable foliage from being interpreted as obstacle. We validate the approach experimentally using sensors mounted on two different mobile robots, navigating in two different environments.