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.

Risk factor analysis and spatiotemporal CART model of cryptosporidiosis in Queensland, Australia

Background: It remains unclear whether it is possible to develop a spatiotemporal epidemic prediction model for cryptosporidiosis disease. This paper examined the impact of social economic and weather factors on cryptosporidiosis and explored the possibility of developing such a model using social economic and weather data in Queensland, Australia. ----- ----- Methods: Data on weather variables, notified cryptosporidiosis cases and social economic factors in Queensland were supplied by the Australian Bureau of Meteorology, Queensland Department of Health, and Australian Bureau of Statistics, respectively. Three-stage spatiotemporal classification and regression tree (CART) models were developed to examine the association between social economic and weather factors and monthly incidence of cryptosporidiosis in Queensland, Australia. The spatiotemporal CART model was used for predicting the outbreak of cryptosporidiosis in Queensland, Australia. ----- ----- Results: The results of the classification tree model (with incidence rates defined as binary presence/absence) showed that there was an 87% chance of an occurrence of cryptosporidiosis in a local government area (LGA) if the socio-economic index for the area (SEIFA) exceeded 1021, while the results of regression tree model (based on non-zero incidence rates) show when SEIFA was between 892 and 945, and temperature exceeded 32°C, the relative risk (RR) of cryptosporidiosis was 3.9 (mean morbidity: 390.6/100,000, standard deviation (SD): 310.5), compared to monthly average incidence of cryptosporidiosis. When SEIFA was less than 892 the RR of cryptosporidiosis was 4.3 (mean morbidity: 426.8/100,000, SD: 319.2). A prediction map for the cryptosporidiosis outbreak was made according to the outputs of spatiotemporal CART models. ----- ----- Conclusions: The results of this study suggest that spatiotemporal CART models based on social economic and weather variables can be used for predicting the outbreak of cryptosporidiosis in Queensland, Australia.

Improving cyclodiode - a risk factor analysis of complications

PURPOSE: The purpose of this study is to identify risk factors for developing complications following treatment of refractory glaucoma with transscleral diode laser cyclophotocoagulation (cyclodiode), to improve the safety profile of this treatment modality. METHOD: A retrospective analysis of 72 eyes from 70 patients who were treated with cyclodiode. RESULTS: The mean pre-treatment IOP was 37.0 mmHg (SD 11.0), with a mean post-treatment reduction in intraocular pressure (IOP) of 19.8 mmHg, and a mean IOP at last follow-up of 17.1 mmHg (SD 9.7). Mean total power delivered during treatment was 156.8 Joules (SD 82.7) over a mean of 1.3 treatments (SD 0.6). Sixteen eyes (22.2% of patients) developed complications from the treatment, with the most common being hypotony, occurring in 6 patients, including 4 with neovascular glaucoma. A higher pre-treatment IOP and higher mean total power delivery also were associated with higher complications. CONCLUSIONS: Cyclodiode is an effective treatment option for glaucoma that is refractory to other treatment options. By identifying risk factors for potential complications, cyclodiode can be modified accordingly for each patient to improve safety and efficacy.

Is diurnal temperature range a risk factor for childhood diarrhea?

Background Previous studies have found that high and cold temperatures increase the risk of childhood diarrhea. However, little is known about whether the within-day variation of temperature has any effect on childhood diarrhea. Methods A Poisson generalized linear regression model combined with a distributed lag non-linear model was used to examine the relationship between diurnal temperature range and emergency department admissions for diarrhea among children under five years in Brisbane, from 1st January 2003 to 31st December 2009. Results There was a statistically significant relationship between diurnal temperature range and childhood diarrhea. The effect of diurnal temperature range on childhood diarrhea was the greatest at one day lag, with a 3% (95% confidence interval: 2%–5%) increase of emergency department admissions per 1°C increment of diurnal temperature range. Conclusion Within-day variation of temperature appeared to be a risk factor for childhood diarrhea. The incidence of childhood diarrhea may increase if climate variability increases as predicted.

Lifestyle risk factor modification in midlife women with type 2 diabetes : theoretical modelling of perceived barriers

Objective: The aim of this paper is to propose a ‘Perceived barriers and lifestyle risk factor modification model’ that could be incorporated into existing frameworks for diabetes education to enhance lifestyle risk factor education in women. Setting: Diabetes education, community health. Primary argument: ‘Perceived barriers’ is a health promotion concept that has been found to be a significant predictor of health promotion behaviour. There is evidence that women face a range of perceived barriers that prevent them from engaging in healthy lifestyle activities. Despite this, current evidence based models of diabetes education do not explicitly incorporate the concept of perceived barriers. A model of risk factor reduction that incorporates ‘perceived barriers’ is proposed. Conclusion: Although further research is required, current approaches to risk factor reduction in type 2 diabetes could be enhanced by identification and goal setting to reduce an individual’s perceived barriers.

Reducing road related injuries for young adolescents : an investigation of truancy as a risk factor

Truancy is recognised as an indicator of engagement in high-risk behaviours for adolescents. Injuries from road related risk behaviours continue to be a leading cause of death and disability for early adolescents (13-14 years). The aim of this research is to determine the extent to which truancy relates to increased risk of road related injuries for early adolescents. Four hundred and twenty-seven Year 9 students (13-14 years) from five high schools in Queensland, Australia, completed a questionnaire about their perceptions of risk and recent injury experience. Self-reported injuries were assessed by the Extended Adolescent Injury Checklist (E-AIC). Injuries resulting from motorcycle use, bicycle use, vehicle use (as passenger or driver), and as a pedestrian were measured for the preceding three months. Students were also asked to indicate whether they sought medical attention for their injuries. Truancy rates were assessed from self-reported skipping class or wagging school over the same three month period. The findings explore the relationship between early adolescent truancy and road related injuries. The relationship between road related injuries and truancy was analysed separately for males and females. Results of this study revealed that road related injuries and reports of associated medical treatment are higher for young people who engage in truancy when compared with non-truant adolescents. The results of this study contribute knowledge about truancy as a risk factor for engagement in road related risks. The findings have the potential to enhance school policies and injury prevention programs if emphasis is placed on increasing school attendance as a safety measure to decrease road related injuries for young adolescents.

Acute illness is a risk factor for impaired respiratory function during nurse-administered procedural sedation and analgesia in the cardiac catheterization laboratory

Impaired respiratory function (IRF) during procedural sedation and analgesia (PSA) poses considerable risk to patient safety as it can lead to inadequate oxygenation and ventilation. Risk factors that can be screened prior to the procedure have not been identified for the cardiac catheterization laboratory (CCL).

Letter : anxiety score as a risk factor for radial artery vasospasm during radial interventions : a pilot study

I read with interest the article in Angiology that determined the role of anxiety level on radial artery spasm during transradial coronary angiography.1 As the importance of conducting more randomised controlled trials using anxiolytics to define the relation between anxiety and vasospasm was noted by the authors, I offer the following insights for investigators to consider when conducting such research. While previous research has already identified that moderate procedural sedation and opioid analgesia reduces the incidence of vasospasm,2 the identification of risk factors in the present study is hypothesis generating as to how outcomes might be even further improved. It is possible that selectively applying either even more intensive sedation and analgesia or complementary non-pharmacological stress-reducing therapies, such as music therapy or visualisation and attentive behaviour, to patients ‘at-risk’ of vasospasm (women and those with high levels of anxiety prior to the procedure) might lead to even better patient outcomes...

Investigation of the low-density lipoprotein receptor gene and cholesterol as a risk factor for migraine

The Low-Density Lipoprotein Receptor (LDLR) gene is a cell surface receptor that plays an important role in cholesterol homeostasis. We investigated the (TA)n polymorphism in exon 18 of the LDLR gene on chromosome 19p13.2 performing an association analysis in 244 typical migraine-affected patients, 151 suffering from migraine with aura (MA), 96 with migraine without aura (MO) and 244 unaffected controls. The populations consisted of Caucasians only, and controls were age- and sex-matched. The results showed no significant difference between groups for allele frequency distributions of the (TA)n polymorphism even after separation of the migraine-affected individuals into subgroups of MA and MO affected patients. This is in contradiction to Mochi et al. who found a positive association of this variant with MO. Our study discusses possible differences between the two studies and extends this research by investigating circulating cholesterol levels in a migraine-affected population.

The methylentetrahydrofolate reductase gene variant (C677T) as a risk factor for essential hypertension in Caucasians

Essential hypertension (EH) is a common, multifactorial disorder likely to be influenced by multiple genes of modest effect. The methylenetetrahydrofolate reductase (MTHFR) gene C677T mutation is functionally important, being strongly associated with reduced enzyme activity and increased plasma levels of homocysteine. Mild hyperhomocysteinemia is a known risk factor for cardiovascular disease (CVD) and hypothesised also to be involved in hypertension pathophysiology. The present study was performed to determine the prevalence of the 677T mutation in Australian Caucasian patients diagnosed with EH and to test whether the C677T variant is associated with the disorder. A case-control cohort, consisting of 250 EH patients and 250 age, sex and racially matched normotensive controls, were used for the association study. Comparison of C677T allele frequencies revealed a higher proportion of the mutant allele (T) in the EH group (40%) compared to unaffected controls (34%) (p=0.07). Furthermore, genotypic results indicated that the prevalence of the homozygous mutant genotype (T/T) in the affected group was higher than that of controls (14%:10%) (p=0.17). Interestingly, conditional logistic regression showed that the MTHFR C677T mutation conferred a mild, yet significant increase in risk of essential hypertension after adjusting for body mass index (odds ratio=1.57, 95% confidence interval: 1.04-2.37, p=0.03). These findings require further investigation in large independent samples, but suggest that essential hypertension, like CVD, may be mildly influenced by the MTHFR C677T variant.

Melanocortin-1 receptor genotype is a risk factor for basal and squamous cell carcinoma

MC1R gene variants have previously been associated with red hair and fair skin color, moreover skin ultraviolet sensitivity and a strong association with melanoma has been demonstrated for three variant alleles that are active in influencing pigmentation: Arg151Cys, Arg160Trp, and Asp294His. This study has confirmed these pigmentary associations with MC1R genotype in a collection of 220 individuals drawn from the Nambour community in Queensland, Australia, 111 of whom were at high risk and 109 at low risk of basal cell carcinoma and squamous cell carcinoma. Comparative allele frequencies for nine MC1R variants that have been reported in the Caucasian population were determined for these two groups, and an association between prevalence of basal cell carcinoma, squamous cell carcinoma, solar keratosis and the same three active MC1R variant alleles was demonstrated [odds ratio = 3.15 95% CI (1.7, 5.82)]. Three other commonly occurring variant alleles: Val60Leu, Val92Met, and Arg163Gln were identified as having a minimal impact on pigmentation phenotype as well as basal cell carcinoma and squamous cell carcinoma risk. A significant heterozygote effect was demonstrated where individuals carrying a single MC1R variant allele were more likely to have fair and sun sensitive skin as well as carriage of a solar lesion when compared with those individuals with a consensus MC1R genotype. After adjusting for the effects of pigmentation on the association between MC1R variant alleles and basal cell carcinoma and squamous cell carcinoma risk, the association persisted, confirming that presence of at least one variant allele remains informative in terms of predicting risk for developing a solar-induced skin lesion beyond that information wained through observation of pigmentation phenotype.