Filter design using radial basis function neural network and genetic algorithm for improved operational health monitoring

The problem of denoising damage indicator signals for improved operational health monitoring of systems is addressed by applying soft computing methods to design filters. Since measured data in operational settings is contaminated with noise and outliers, pattern recognition algorithms for fault detection and isolation can give false alarms. A direct approach to improving the fault detection and isolation is to remove noise and outliers from time series of measured data or damage indicators before performing fault detection and isolation. Many popular signal-processing approaches do not work well with damage indicator signals, which can contain sudden changes due to abrupt faults and non-Gaussian outliers. Signal-processing algorithms based on radial basis function (RBF) neural network and weighted recursive median (WRM) filters are explored for denoising simulated time series. The RBF neural network filter is developed using a K-means clustering algorithm and is much less computationally expensive to develop than feedforward neural networks trained using backpropagation. The nonlinear multimodal integer-programming problem of selecting optimal integer weights of the WRM filter is solved using genetic algorithm. Numerical results are obtained for helicopter rotor structural damage indicators based on simulated frequencies. Test signals consider low order polynomial growth of damage indicators with time to simulate gradual or incipient faults and step changes in the signal to simulate abrupt faults. Noise and outliers are added to the test signals. The WRM and RBF filters result in a noise reduction of 54 - 71 and 59 - 73% for the test signals considered in this study, respectively. Their performance is much better than the moving average FIR filter, which causes significant feature distortion and has poor outlier removal capabilities and shows the potential of soft computing methods for specific signal-processing applications.

Genetic Heterogeneity in Wild Isolates of Cellular Slime Mold Social Groups

This study addresses the issues of spatial distribution, dispersal, and genetic heterogeneity in social groups of the cellular slime molds (CSMs). The CSMs are soil amoebae with an unusual life cycle that consists of alternating solitary and social phases. Because the social phase involves division of labor with what appears to be an extreme form of "altruism", the CSMs raise interesting evolutionary questions regarding the origin and maintenance of sociality. Knowledge of the genetic structure of social groups in the wild is necessary for answering these questions. We confirm that CSMs are widespread in undisturbed forest soil from South India. They are dispersed over long distances via the dung of a variety of large mammals. Consistent with this mode of dispersal, most social groups in the two species examined for detailed study, Dictyostelium giganteum and Dictyostelium purpureum, are multi-clonal.

Cognitive Functioning in Alcohol and Other Substance Use Disorders in Young Adulthood : A Genetic Epidemiological Study

Alcohol and other substance use disorders (SUDs) result in great costs and suffering for individuals and families and constitute a notable public health burden. A multitude of factors, ranging from biological to societal, are associated with elevated risk of SUDs, but at the level of individuals, one of the best predictors is a family history of SUDs. Genetically informative twin and family studies have consistently indicated this familial risk to be mainly genetic. In addition, behavioral and temperamental factors such as early initiation of substance use and aggressiveness are associated with the development of SUDs. These familial, behavioral and temperamental risk factors often co-occur, but their relative importance is not well known. People with SUDs have also been found to differ from healthy controls in various domains of cognitive functioning, with poorer verbal ability being among the most consistent findings. However, representative population-based samples have rarely been used in neuropsychological studies of SUDs. In addition, both SUDs and cognitive abilities are influenced by genetic factors, but whether the co-variation of these traits might be partly explained by overlapping genetic influences has not been studied. Problematic substance use also often co-occurs with low educational level, but it is not known whether these outcomes share part of their underlying genetic influences. In addition, educational level may moderate the genetic etiology of alcohol problems, but gene-environment interactions between these phenomena have also not been widely studied. The incidence of SUDs peaks in young adulthood rendering epidemiological studies in this age group informative. This thesis investigated cognitive functioning and other correlates of SUDs in young adulthood in two representative population-based samples of young Finnish adults, one of which consisted of monozygotic and dizygotic twin pairs enabling genetically informative analyses. Using data from the population-based Mental Health in Early Adulthood in Finland (MEAF) study (n=605), the lifetime prevalence of DSM-IV any substance dependence or abuse among persons aged 21—35 years was found to be approximately 14%, with a majority of the diagnoses being alcohol use disorders. Several correlates representing the domains of behavioral and affective factors, parental factors, early initiation of substance use, and educational factors were individually associated with SUDs. The associations between behavioral and affective factors (attention or behavior problems at school, aggression, anxiousness) and SUDs were found to be largely independent of factors from other domains, whereas daily smoking and low education were still associated with SUDs after adjustment for behavioral and affective factors. Using a wide array of neuropsychological tests in the MEAF sample and in a subsample (n=602) of the population-based FinnTwin16 (FT16) study, consistent evidence of poorer verbal cognitive ability related to SUDs was found. In addition, participants with SUDs performed worse than those without disorders in a task assessing psychomotor processing speed in the MEAF sample, whereas no evidence of more specific cognitive deficits was found in either sample. Biometrical structural equation models of the twin data suggested that both alcohol problems and verbal ability had moderate heritabilities (0.54—0.72), and that their covariation could be explained by correlated genetic influences (genetic correlations -0.20 to -0.31). The relationship between educational level and alcohol problems, studied in the full epidemiological FT16 sample (n=4,858), was found to reflect both genetic correlation and gene-environment interaction. The co-occurrence of low education and alcohol problems was influenced by overlapping genetic factors. In addition, higher educational level was associated with increased relative importance of genetic influences on alcohol problems, whereas environmental influences played a more important role in young adults with lower education. In conclusion, SUDs, especially alcohol abuse and dependence, are common among young Finnish adults. Behavioral and affective factors are robustly related to SUDs independently of many other factors, and compared to healthy peers, young adults who have had SUDs during their life exhibit significantly poorer verbal cognitive ability, and possibly less efficient psychomotor processing. Genetic differences between individuals explain a notable proportion of individual differences in risk of alcohol dependence, verbal ability, and educational level, and the co-occurrence of alcohol problems with poorer verbal cognition and low education is influenced by shared genetic backgrounds. Finally, various environmental factors related to educational level in young adulthood moderate the relative importance of genetic factors influencing the risk of alcohol problems, possibly reflecting differences in social control mechanisms related to educational level.

Optimal reservoir operation for irrigation of multiple crops using genetic algorithms

This paper presents a genetic algorithm (GA) model for obtaining an optimal operating policy and optimal crop water allocations from an irrigation reservoir. The objective is to maximize the sum of the relative yields from all crops in the irrigated area. The model takes into account reservoir inflow, rainfall on the irrigated area, intraseasonal competition for water among multiple crops, the soil moisture dynamics in each cropped area, the heterogeneous nature of soils. and crop response to the level of irrigation applied. The model is applied to the Malaprabha single-purpose irrigation reservoir in Karnataka State, India. The optimal operating policy obtained using the GA is similar to that obtained by linear programming. This model can be used for optimal utilization of the available water resources of any reservoir system to obtain maximum benefits.

Filter design using radial basis function neural network and genetic algorithm for improved operational health monitoring

The problem of denoising damage indicator signals for improved operational health monitoring of systems is addressed by applying soft computing methods to design filters. Since measured data in operational settings is contaminated with noise and outliers, pattern recognition algorithms for fault detection and isolation can give false alarms. A direct approach to improving the fault detection and isolation is to remove noise and outliers from time series of measured data or damage indicators before performing fault detection and isolation. Many popular signal-processing approaches do not work well with damage indicator signals, which can contain sudden changes due to abrupt faults and non-Gaussian outliers. Signal-processing algorithms based on radial basis function (RBF) neural network and weighted recursive median (WRM) filters are explored for denoising simulated time series. The RBF neural network filter is developed using a K-means clustering algorithm and is much less computationally expensive to develop than feedforward neural networks trained using backpropagation. The nonlinear multimodal integer-programming problem of selecting optimal integer weights of the WRM filter is solved using genetic algorithm. Numerical results are obtained for helicopter rotor structural damage indicators based on simulated frequencies. Test signals consider low order polynomial growth of damage indicators with time to simulate gradual or incipient faults and step changes in the signal to simulate abrupt faults. Noise and outliers are added to the test signals. The WRM and RBF filters result in a noise reduction of 54 - 71 and 59 - 73% for the test signals considered in this study, respectively. Their performance is much better than the moving average FIR filter, which causes significant feature distortion and has poor outlier removal capabilities and shows the potential of soft computing methods for specific signal-processing applications. (C) 2005 Elsevier B. V. All rights reserved.

Determinants of Cognitive Impairment in Old Age and Their Genetic Architecture

Cognitive health is of central importance for independent and balanced old age, while memory disorders represent the leading cause of intensive and long-term care among the Finnish elderly. The aims of this study were to analyse the effect of height, body mass index, weight change, metabolic conditions and coffee drinking in midlife on cognitive performance in old age among a sample of 2606 Finnish twins aged 65 years or older who had participated in a telephone interview to assess their cognitive status. Since coffee drinking associates with several metabolic conditions and Finns are known to be the greatest consumers of coffee in the world, the heritability and stability of coffee drinking was analysed in the whole Older Finnish Twin Cohort (n=10716). In order to investigate the association between height and cognitive performance in a population with more supportive childhood living conditions, a total of 2161 Danish twins were included in this study. A greater height was found to clearly associate with better cognitive performance in Finnish subjects, but less so among the Danish sample, which may reflect the childhood environmental differences between these cohorts. In the Finnish subjects, there was greater variance in cognitive performance among shorter subjects, and environmental factors were found to play a greater role in their cognitive performance, whereas the cognitive performance of taller participants was mainly explained by genetic factors. Midlife metabolic variables that were found to be significantly associated with a poorer cognitive performance in old age included a higher body mass index and three metabolic conditions: cardiovascular disease, hypertension and, most significantly of all, diabetes. Moreover, both weight gain and loss, even to a lesser degree than suggested previously, were found to be associated with poorer cognition. Furthermore, evidence of a causal relationship between midlife cardiovascular disease and cognitive performance in old age was demonstrated among discordant twin pairs. Conversely, no effect of coffee drinking in midlife on cognitive performance in old age was observed, although coffee drinking was demonstrated to be stable in the study population. The heritability of coffee drinking was found to differ across sexes and age groups, being 51% in men and 52% in women in the whole study population. This study supports the contention that cognitive performance in old age reflects the effects of multiple genetic and environmental exposures, including their complex interactions during the life-span. The demonstrated associations and evidence of a causal pathway between potentially preventable exposures and poorer cognitive performance highlight the importance of preventive medicine.

Generating artificial chromosomes with probability control in genetic algorithm for machine scheduling problems

In this paper, a novel genetic algorithm is developed by generating artificial chromosomes with probability control to solve the machine scheduling problems. Generating artificial chromosomes for Genetic Algorithm (ACGA) is closely related to Evolutionary Algorithms Based on Probabilistic Models (EAPM). The artificial chromosomes are generated by a probability model that extracts the gene information from current population. ACGA is considered as a hybrid algorithm because both the conventional genetic operators and a probability model are integrated. The ACGA proposed in this paper, further employs the ``evaporation concept'' applied in Ant Colony Optimization (ACO) to solve the permutation flowshop problem. The ``evaporation concept'' is used to reduce the effect of past experience and to explore new alternative solutions. In this paper, we propose three different methods for the probability of evaporation. This probability of evaporation is applied as soon as a job is assigned to a position in the permutation flowshop problem. Experimental results show that our ACGA with the evaporation concept gives better performance than some algorithms in the literature.

Genetic and epigenetic mechanisms of tumor predisposition in hereditary non-polyposis colorectal carcinoma and sporadic cancers

Individuals with inherited deficiency in DNA mismatch repair(MMR) (Lynch syndrome) LS are predisposed to different cancers in a non-random fashion. Endometrial cancer (EC) is the most common extracolonic malignancy in LS. LS represents the best characterized form of hereditary nonpolyposis colorectal carcinoma (HNPCC). Other forms of familial non-polyposis colon cancer exist, including familial colorectal cancer type X (FCCX). This syndrome resembles LS, but MMR gene defects are excluded and the predisposition genes are unknown so far. To address why different organs are differently susceptible to cancer development, we examined molecular similarities and differences in selected cancers whose frequency varies in LS individuals. Tumors that are common (colorectal, endometrial, gastric) and less common (brain, urological) in LS were characterized for MMR protein expression, microsatellite instability (MSI), and by altered DNA methylation. We also studied samples of histologically normal endometrium, endometrial hyperplasia,and cancer for molecular alterations to identify potential markers that could predict malignant transformation in LS and sporadic cases. Our results suggest that brain and kidney tumors follow a different pathway for cancer development than the most common LS related cancers.Our results suggest also that MMR defects are detectable in endometrial tissues from a proportion of LS mutation carriers prior to endometrial cancer development. Traditionally (complex) atypical hyperplasia has been considered critical for progression to malignancy. Our results suggest that complex hyperplasia without atypia is equally important as a precursor lesion of malignancy. Tumor profiles from Egypt were compared with colorectal tumors from Finland to evaluate if there are differences specific to the ethnic origin (East vs.West). Results showed for the first time a distinct genetic and epigenetic signature in the Egyptian CRC marked by high methylation of microsatellite stable tumors associated with advanced stage, and low frequency of Wnt signaling activation, suggesting a novel pathway. DNA samples from FCCX families were studied with genome wide linkage analysis using microsatellite markers. Selected genes from the linked areas were tested for possible mutations that could explain predisposition to a large number of colon adenomas and carcinomas seen in these families. Based on the results from the linkage analysis, a number of areas with tentative linkage were identified in family 20. We narrowed down these areas by additional microsatellite markers to found a mutation in the BMPR1A gene. Sequencing of an additional 17 FCCX families resulted in a BMPR1A mutation frequency of 2/18 families (11%). Clarification of the mechanisms of the differential tumor susceptibility in LS increases the understanding of gene and organ specific targets of MMR deficiency. While it is generally accepted that widespread MMR deficiency and consequent microsatellite instability (MSI) drives tumorigenesis in LS, the timing of molecular alterations is controversial. In particular, it is important to know that alterations may occur several years before cancer formation, at stages that are still histologically regarded as normal. Identification of molecular markers that could predict the risk of malignant transformation may be used to improve surveillance and cancer prevention in genetically predisposed individuals. Significant fractions of families with colorectal and/or endometrial cancer presently lack molecular definition altogether. Our findings expand the phenotypic spectrum of BMPR1A mutations and, for the first time, link FCCX families to the germline mutation of a specific gene. In particular, our observations encourage screening of additional families with FCCX for BMPR1A mutation, which is necessary in obtaining a reliable estimate of the share of BMPR1A-associated cases among all FCCX families worldwide. Clinically, the identification of predisposing mutations enables targeted cancer prevention in proven mutation carriers and thereby reduces cancer morbidity and mortality in the respective families.

Microcycle conidiation and its genetic basis in Neurospora crassa

Some wild isolates of Neurospora show microcycle conidiation in liquid culture under continuous agitation. Macroconidia from agar-grown mycelial cultures germinated in liquid and the germlings spontaneously produced conidia with no intervening mycelial phase. Three types of microcycle conidiation were seen among progeny of N. crassa Vickramam A x N. crassa a wild-type: (1) multinucleate blastoconidia produced by apical budding and septation, (2) multinucleate arthroconidia produced by holothallic septation and disarticulation of cells, and (3) uninucleate microconidia produced directly from conidiogenous cells of the germlings. Two genes were identified which control specific patterns of microcycle conidiogenesis. A single gene mcb in linkage group VR near al-3 (3.2% recombination) controls blastoconidiation. This gene is epistatic to gene mcm located in linkage group IIL, very near ro-7 (1.4%). mcm controls both microconidiation and arthroconidiation depending on temperature. Strains of genotype mcm produce microconidia almost exclusively at 18-22 degrees C, but arthroconidia with few or no microconidia at 30 degrees C. Because they result in rapid and synchronized conidiation in liquid culture, the two genes should be useful for studies of developmental gene regulation. mcm makes it possible to obtain large quantities of pure microconidia rapidly for experimentation.

Genetic polymorphisms and laboratory variables as predictors of blood pressure response to antihypertensive drugs

Hypertension is one of the major risk factors for cardiovascular morbidity. The advantages of antihypertensive therapy have been clearly demonstrated, but only about 30% of hypertensive patients have their blood pressure (BP) controlled by such treatment. One of the reasons for this poor BP control may lie in the difficulty in predicting BP response to antihypertensive treatment. The average BP reduction achieved is similar for each drug in the main classes of antihypertensive agents, but there is a marked individual variation in BP responses to any given drug. The purpose of the present study was to examine BP response to four different antihypertensive monotherapies with regard to demographic characteristics, laboratory test results and common genetic polymorphisms. The subjects of the present study are participants in the pharmacogenetic GENRES Study. A total of 208 subjects completed the whole study protocol including four drug treatment periods of four weeks, separated by four-week placebo periods. The study drugs were amlodipine, bisoprolol, hydrochlorothiazide and losartan. Both office (OBP) and 24-hour ambulatory blood pressure (ABP) measurements were carried out. BP response to study drugs were related to basic clinical characteristics, pretreatment laboratory test results and common polymorphisms in genes coding for components of the renin-angiotensin system, alpha-adducin (ADD1), beta1-adrenergic receptor (ADRB1) and beta2-adrenergic receptor (ADRB2). Age was positively correlated with BP responses to amlodipine and with OBP and systolic ABP responses to hydrochlorothiazide, while body mass index was negatively correlated with ABP responses to amlodipine. Of the laboratory test results, plasma renin activity (PRA) correlated positively with BP responses to losartan, with ABP responses to bisoprolol, and negatively with ABP responses to hydrochlorothiazide. Uniquely to this study, it was found that serum total calcium level was negatively correlated with BP responses to amlodipine, whilst serum total cholesterol level was negatively correlated with ABP responses to amlodipine. There were no significant associations of angiotensin II type I receptor 1166A/C, angiotensin converting enzyme I/D, angiotensinogen Met235Thr, ADD1 Gly460Trp, ADRB1 Ser49Gly and Gly389Arg and ADRB2 Arg16Gly and Gln27Glu polymorphisms with BP responses to the study drugs. In conclusion, this study confirmed the relationship between pretreatment PRA levels and response to three classes of antihypertensive drugs. This study is the first to note a significant inverse relation between serum calcium level and responsiveness to a calcium channel blocker. However, this study could not replicate the observations that common polymorphisms in angiotensin II type I receptor, angiotensin converting enzyme, angiotensinogen, ADD1, ADRB1, or ADRB2 genes can predict BP response to antihypertensive drugs.

Hostility: A prospective study of the genetic and environmental background and associations with cardiovascular risk

Hostility is a multidimensional construct having wide effects on society. In its different forms, hostility is related to a large array of social and health problems, such as criminality, substance abuse, depression, and cardiovascular risks. Identifying and tackling early-life factors that contribute to hostility may have public health significance. Although the variance in hostility is estimated to be 18-50 percent heritable, there are significant gaps in knowledge regarding the molecular genetics of hostility. It is known that a cold and unsupportive home atmosphere in childhood predicts a child s later hostility. However, the long-term effects of care-giving quality on hostility in adulthood and the role of genes in this association are unclear. The present dissertation is part of the ongoing population-based prospective Young Finns study, which commenced in 1980 with 3596 3-18-year-old boys and girls who were followed for 27 years. The specific aims of the dissertation were first to study the antecedents of hostility by looking at 1) the genetic background, 2) the early environmental predictors, and 3) the gene environment interplay behind hostility. As a second aim, the thesis endeavored to examine 4) the association between hostility and cardiovascular risks, and 5) the moderating effect of demographic factors, such as gender and socioeconomic status, on this association. The study found potential gene polymorphisms from chromosomes 7, 14, 17, and 22 suggestively associated with hostility. Of early environmental influences, breastfeeding and early care-giving were found to predict hostility in adulthood. In addition, a serotonin receptor 2A polymorphism rs6313 moderated the effect of early care-giving on later hostile attitudes. Furthermore, hostility was shown to predict cardiovascular risks, such as metabolic syndrome and inflammation. Finally, parental socioeconomic status was found to moderate the association between anger and early atherosclerosis. The new genetic and early environmental antecedents of hostility identified in this research may help in understanding the development of hostility and its health risks, and in planning appropriate prevention. The significance of early influences on this development is stressed. Although the markers studied are individual- and family-related factors, these may be influenced at the societal level by giving accurate information to all individuals concerned and by improving the societal circumstances.

A Genetic Study of Immunoglobulin E and Atopic Disease Based on Families Ascertained through Asthmatic Children

In order to investigate the modes of inheritance of serum immunoglobulin E (IgE) levels and atopic disease, serum IgE levels and data on allergic disease were obtained from 42 families ascertained through asthmatic children visiting an allergy clinic. Although the mean IgE levels were elevated (mean 637 U/ml), the prevalence of atopic disease in this population was surprisingly low. When the data were analyzed using complex segregation analysis, no major locus could be detected. Moreover, the polygenic heritability was unexpectedly small even though the correlation between serum IgE levels and the liability to atopic disease was around 0.4. Given this unusual set of findings, it is postulated that parasitic infections in this population have (in accordance with well-established results of parasitic disease) caused both elevated levels of serum IgE and a decreased prevalence of allergic disease with the possible masking of the various genetic components of serum IgE levels and atopic disease.

Transposable element-medicated evolution of sex: A population genetic model

For a population made up of individuals capable of sexual as well as asexual modes of reproduction, conditions for the spread of a transposable element are explored using a one-locus, two-haplotype model. The analysis is then extended to include the possibility that the transposable element can modulate the probability of sexual reproduction, thus casting Hickey’s (1982,Genetics 101: 519–531) suggestion in a population genetics framework. The model explicitly includes the cost of sexual reproduction, fitness disadvantage to the transposable element, probability of transposition, and the predisposition for sexual reproduction in the presence and absence of the transposable element. The model predicts several kinds of outcome, including initial frequency dependence and stable polymorphism. More importantly, it is seen that for a wide range of parameter values, the transposable element can go to fixation. Therefore it is able to convert the population from a predominantly asexual to a predominantly sexual mode of reproduction. Viewed in conjunction with recent results implicating short stretches of apparently non-coding DNA in sex determination (McCoubreyet al. 1988,Science 242: 1146–1151), the model hints at the important role this mechanism could have played in the evolution of sexuality.

A near-optimal initial seed value selection in K-means means algorithm using a genetic algorithm

The K-means algorithm for clustering is very much dependent on the initial seed values. We use a genetic algorithm to find a near-optimal partitioning of the given data set by selecting proper initial seed values in the K-means algorithm. Results obtained are very encouraging and in most of the cases, on data sets having well separated clusters, the proposed scheme reached a global minimum.