Backward Elimination Technique Research Articles

An efficient adaptive forward–backward selection method for sparse polynomial chaos expansion

As an efficient uncertainty quantification (UQ) methodology for moment propagation and probability analysis of quantities of interest, polynomial chaos (PC) expansions have received broad and sustained attentions. However, the exponentially increasing cost of building PC representations with increasing dimension of uncertainty, i.e., the curse of dimensionality, seriously restricts the practical application of PC at the industrial level. Some efficient strategies applying adaptive basis selection algorithm for sparse optimization (or l1-minimization) of PC show great potential compared to the classical full PC. However, these strategies mainly focus on forward selection algorithms, which are incapable of correcting any error made by these algorithms. Hence, this paper develops a novel adaptive forward–backward selection (AFBS) algorithm for reconstructing sparse PC. The proposed algorithm by a reasonable combination of forward selection and adaptively backward elimination technique can efficiently correct mistakes made by earlier forward selection steps, which retains the most significant PC terms and discards redundant or insignificant ones. The accuracy of built PC metamodel is checked by a cross-validation procedure. As a consequence, the most significant PC terms are detected sequentially and corresponding PC coefficients are accurately recovered. It largely enhances the sparsity of PC and improves the prediction accuracy compared to the popular forward selection algorithms, e.g., least angle regressions (LARs). To validate the efficiency of the proposed algorithm, a complex analytical function with Gaussian distribution inputs and two challenging aerodynamic applications including a sonic boom propagation analysis considering atmospheric uncertainty and a natural-laminar-flow (NLF) airfoil computation under both geometrical and operational uncertainties are elaborated. With an in-depth comparison with some popular PC reconstruction methodologies, the performance of the devised AFBS method is assessed comprehensively. The results demonstrate that the proposed AFBS method can efficiently identify the significant PC contributions describing the problems, and reproduce sparser PC metamodel and more accurate UQ results than those classical full PC and LARs methods.

Relative contributions of auditory and cognitive functions on speech recognition in quiet and in noise among older adults

IntroductionHearing acuity, central auditory processing and cognition contribute to the speech recognition difficulty experienced by older adults. Therefore, quantifying the contribution of these factors on speech recognition problem is important in order to formulate a holistic and effective rehabilitation. ObjectiveTo examine the relative contributions of auditory functioning and cognition status to speech recognition in quiet and in noise. MethodsWe measured speech recognition in quiet and in composite noise using the Malay Hearing in noise test on 72 native Malay speakers (60–82 years) older adults with normal to mild hearing loss. Auditory function included pure tone audiogram, gaps-in-noise, and dichotic digit tests. Cognitive function was assessed using the Malay Montreal cognitive assessment. ResultsLinear regression analyses using backward elimination technique revealed that had the better ear four frequency average (0.5–4kHz) (4FA), high frequency average and Malay Montreal cognitive assessment attributed to speech perception in quiet (total r2=0.499). On the other hand, high frequency average, Malay Montreal cognitive assessment and dichotic digit tests contributed significantly to speech recognition in noise (total r2=0.307). Whereas the better ear high frequency average primarily measured the speech recognition in quiet, the speech recognition in noise was mainly measured by cognitive function. ConclusionsThese findings highlight the fact that besides hearing sensitivity, cognition plays an important role in speech recognition ability among older adults, especially in noisy environments. Therefore, in addition to hearing aids, rehabilitation, which trains cognition, may have a role in improving speech recognition in noise ability of older adults.

Applied statistical functions and multivariate analysis of geochemical compositional data to evaluate mineralization in Glojeh polymetallic deposit, NW Iran

Various genesis of epithermal veins as well as host rock cause complication in the modeling process. Thus LINEST and controlling function were applied to improve the accuracy and the quality of the model. The LINEST is a model which is based on multiple linear regression and refers to a branch of applied statistics. This method concerns directly to the application of t-test (TINV and TDIST to analyses of variables in the model) and F-test (FDIST, F-statistic to compare different models) analysis. Backward elimination technique is applied to reduce the number of variables in the model through all the borehole data. After 18 steps, an optimized reduced model (ORM) was constructed and ranked in order of importance as Pb>Ag>P>Hg>Mn>Nb>U>Sr>Sn>As>Cu, with the lowest confidence level (CL) of 92% for Cu. According to the epigenetic vein genesis of Glojeh polymetallic deposit, determination of spatial patterns and elemental associations accompanied by anomaly separation were conducted by K-means cluster and robust factor analysis method based on centered log-ratio (clr) transformed data. Therefore, 12 samples (cluster 2) with the maximum distance from centroid, indicates the intensity of vein polymetallic mineralization in the deposit. In addition, an ORM for vein population was extracted for Sb>Al>As>Mg>Pb>Cu>Ag elements with the R2 up to 0.99. On the other hand, after 23 steps of optimization process at the host rock population, an ORM was conducted by Ag>Te>Hg>Pb>Mg>Al>Sb>As represented in descending order of t-values. It revealed that Te and Hg can be considered as pathfinder elements for Au at the host rock. Based on the ORMs at each population Ag, Pb, and As were often associated with Au mineralization. The concentration ratio of (tSb × tAl)vein/(tSb × tAl)background as an enrichment index can intensify the mineralization detection. Finally, Glojeh deposit was evaluated to be classified as a vein-style Au (Ag, Pb, As)-polymetallic mineralization.

PLATELET TRANSFUSION PREDICTION IN CARDIAC SURGERY

Impaired platelet function from cardiopulmonary bypass and the use of preoperative anti-platelet agents are important factors in cardiac surgery blood loss. Platelet transfusions are commonly used in the postoperative phase to mitigate bleeding, but subject’s patients to the risk of infection, alloimmunization and lung injury. Guidelines exist for platelet administration; however, there is significant transfusion variability amongst centers. The lack of evidence based criteria for the administration of platelet transfusion in cardiac surgery creates this heterogeneity. In order to enhance our understanding of platelet transfusion, we intended to identify clinical risk factors associated with platelet transfusion and to subsequently develop a platelet transfusion prediction model for cardiac surgical procedures. The Cardiac Surgery Blood Utilization Database (CSBUD) is our hospitals proprietary software used to capture demographic and transfusion specific data for each patient undergoing cardiac surgery. The CSBUD was interrogated (2013-2018) for all consecutive patients (n=2648). The data was randomly divided into a development set (n=1661) and validation set (n=987). Univariable analysis was used to identify covariates for model inclusion (p < 0.25). A multivariable logistic regression model was used to assess treatment effects. A supervised backward stepwise elimination technique (p < 0.10) was employed to identify independent predictors for the final model. The independent predictor coefficients were used to derive a platelet transfusion risk score. The analysis identified preoperative platelet count (OR=0.985, 95%CI: 0.981-0.989, p<0.001), ADP inhibitor (OR=2.507, 95% CI: 1.064-5.904, p=0.036) and GPIIb/IIIa (OR=8.915, 95% CI: 1.061-74.859, p=0.044) use; emergent surgery (OR=1.179, 95% CI: 1.079-2.411, p<0.001), aortic surgery (OR=7.349, 95% CI: 4.017-13.445, p<0.001) and RBC transfusion (OR=15.44, 95% CI: 9.736-24.497, p<0.001) as being independently associated with platelet transfusion. The risk score was divided into low, medium and high categories with a combined (development set and validation set) transfusion rate of 3%, 32% and 86%, respectively (Table 1). No significant differences existed between the development and validation sets. The model demonstrated excellent discrimination (Figure 1). Our analysis identified six independent clinical factors predictive of platelet transfusions. This enabled us to develop a highly discriminative platelet transfusion prediction model, which classifies the probability of a platelet transfusion into low, medium and high-risk categories. To our knowledge this is the first report of such a predictive model and we believe this will help clinicians to tailor their perioperative management accordingly.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

The effect of knowledge on risk perception for Hepatitis b infection among housekeepers in a tertiary hospital in Nigeria

Background: Risk perceptions are central to many theories of health behavior and consistent relationships have been demonstrated between risk perceptions and relevant health behaviors (1, 2). Studies have identified knowledge and prior screening as potent predictors of risk perception for Hepatitis B infection among doctors and nurses (3). Housekeepers, like other health workers, have occupational exposure to hepatitis B infection but are less often studied and are thus often overlooked during control efforts for occupational exposures (4). The aim of this study was to identify potent predictors of risk perception for hepatitis B infection among housekeepers. Methods & Materials: A cross-sectional hospital-based study of 89 consenting hospital cleaners was conducted in Nigeria. A 58-item validated questionnaire (Cronbach α = 0.82) was used to assess participants’ characteristics, knowledge and risk perception for hepatitis B infection. Multivariate logistic regression using the backward elimination technique was used to assess the effect of knowledge on risk perception for hepatitis B infection while controlling for other covariates (age, sex, work experience, level of education, and prior HBsAg screening). Risk perception was dichotomized to low and high. The level of significance was set at 0.05. Results: The mean (hepatitis B infection) knowledge score was 4.24 ± 3.69 (range = 0-12 out of 26). Majority (89.9%) of the participants had poor knowledge (scores < 9) while the remaining 10.1% had fair knowledge about hepatitis B infection (scores 9-13). Fifty-eight (65.2%), 12 (13.5%), and 19 (21.3%) of the participants reported low, moderate, and high risk perception for hepatitis B infection respectively. The best model for predicting risk perception at P < 0.05) contained only total knowledge score (Pseudo R2 = 10.24). Area under the ROC curve was 70.91%, showing good discrimination. The odds ratio was 1.25 (95% CI = 1.09-1.43). Prior HbsAg screening was not a predictor of risk perception for hepatitis B infection. Conclusion: Risk perception is predicted by knowledge about hepatitis B infection. The housekeepers had poorer knowledge and low risk perception for hepatitis B infection when compared with other health workers. There is a need for targeted strategies to improve housekeepers’ knowledge about hepatitis B infection.

Object-based feature selection for crop classification using multi-temporal high-resolution imagery

ABSTRACTWith high-resolution remote-sensing data, there are numerous possible features for object description, making the selection of optimal features a time-consuming and subjective process. While substantial efforts have been made to compare the utility of feature selection metrics, less attention has been paid to the efficiency of such in the context of object-based image analysis. In this study, the statistical measurement z-score was used to ensure compatibility with objects. We assessed the feasibility of a z-score method, and then ranked and reduced input features using a backward elimination technique. The results showed that separability can be efficiently estimated based on z-score values, and the near-infrared band performed the best for crop classification. A straightforward trend was observed, and the optimal feature set was created, which was a combination of spectral, temporal, texture information and vegetation indices. These features complement one another to help increase crop map accuracy. For the 40% of the entire sample sizes, the optimal feature sets produced the best trade-off between the number of inputs and classification accuracy, with the misclassification error of 7.09%. Additionally, reliable crop maps were obtained, with the overall accuracy of 92.64%, and the z-score method showed great potential for the separability of crops at object scale using remotely sensed multi-temporal data.

Elimination of Backward Browsing using Decomposition and Compression for Efficient Navigation Prediction

Analysing the user's browsing patterns stored in weblog file can help in providing the personalised environment, improving website structure and recommending the suitable navigation pattern. While browsing the web, the user navigates in a forward direction following the web topology, results in the session having correlated web pages. Backward navigation takes place when the user returns to the previously visited page. This traversal results in increased session length, complexity and reduction in the prediction accuracy. The frequent backward movement also infers that the web community is not well structured. Therefore, to discover the meaningful browsing pattern and improve the website structure, it is essential to filter the repeated web pages from the session. This paper proposes two novel backward elimination techniques: decomposed backward browsing (BBDcom) and compressed backward browsing (BBCom). BBDcom reduces the length of a session using decomposition. It can restore the original session without losing data. BBCom compresses the session length by eliminating web page(s) lying in between the redundant web pages. The experimental result shows that proposed techniques improve the prediction accuracy and reduces the state-space complexity.

Elimination of backward browsing using decomposition and compression for efficient navigation prediction

Analysing the user's browsing patterns stored in weblog file can help in providing the personalised environment, improving website structure and recommending the suitable navigation pattern. While browsing the web, the user navigates in a forward direction following the web topology, results in the session having correlated web pages. Backward navigation takes place when the user returns to the previously visited page. This traversal results in increased session length, complexity and reduction in the prediction accuracy. The frequent backward movement also infers that the web community is not well structured. Therefore, to discover the meaningful browsing pattern and improve the website structure, it is essential to filter the repeated web pages from the session. This paper proposes two novel backward elimination techniques: decomposed backward browsing (BBDcom) and compressed backward browsing (BBCom). BBDcom reduces the length of a session using decomposition. It can restore the original session without losing data. BBCom compresses the session length by eliminating web page(s) lying in between the redundant web pages. The experimental result shows that proposed techniques improve the prediction accuracy and reduces the state-space complexity.

Assessing the utility of the spot 6 sensor in detecting and mapping &lt;i&gt;Lantana camara&lt;/i&gt; for a community clearing project in KwaZulu-Natal, South Africa

Lantana camara is a significant weed in South Africa which is causing severe impacts on agriculture by reducing grazing areas. This study assessed the potential of the SPOT 6 multispectral sensor and two broadband vegetation indices (NDVI and SR) for detecting and mapping Lantana camara in a community grazing land in KwaZulu-Natal, South Africa. The SPOT 6 bands and vegetation indices successfully classified Lantana camara with an overall accuracy of 75% on an independent test dataset using the random forest algorithm. Furthermore, it was tested if the random forest model based on variable importance (VIP) could improve the classification accuracy using the best subset of bands and indices. A backward feature elimination technique was used to select the best subset of VIP bands and indices to improve the classification. By eliminating SPOT bands 1 and 4 which yielded the lowest VIP scores the random forest model improved the classification accuracy to 83.33% on an independent test dataset. The study indicates the potential of satellite remote sensing in weed detection and mapping in South Africa using readily available multispectral data to assist poorer communities in grazing management.

Feature selection for position estimation using an omnidirectional camera

This paper considers visual feature selection to implement position estimation using an omnidirectional camera. The localization is based on a maximum likelihood estimation (MLE) with a map from optimally selected visual features using Gaussian process (GP) regression. In particular, the collection of selected features over a surveillance region is modeled by a multivariate GP with unknown hyperparameters. The hyperparameters are identified through the learning process by an MLE, which are used for prediction in an empirical Bayes fashion. To select features, we apply a backward sequential elimination technique in order to improve the quality of the position estimation with compressed features for efficient localization. The excellent results of the proposed algorithm are illustrated by the experimental studies with different visual features under both indoor and outdoor real-world scenarios.

Comparison between WorldView-2 and SPOT-5 images in mapping the bracken fern using the random forest algorithm

Plant species invasion is known to be a major threat to socioeconomic and ecological systems. Due to high cost and limited extents of urban green spaces, high mapping accuracy is necessary to optimize the management of such spaces. We compare the performance of the new-generation WorldView-2 (WV-2) and SPOT-5 images in mapping the bracken fern [Pteridium aquilinum (L) kuhn] in a conserved urban landscape. Using the random forest algorithm, grid-search approaches based on out-of-bag estimate error were used to determine the optimal ntree and mtry combinations. The variable importance and backward feature elimination techniques were further used to determine the influence of the image bands on mapping accuracy. Additionally, the value of the commonly used vegetation indices in enhancing the classification accuracy was tested on the better performing image data. Results show that the performance of the new WV-2 bands was better than that of the traditional bands. Overall classification accuracies of 84.72 and 72.22% were achieved for the WV-2 and SPOT images, respectively. Use of selected indices from the WV-2 bands increased the overall classification accuracy to 91.67%. The findings in this study show the suitability of the new generation in mapping the bracken fern within the often vulnerable urban natural vegetation cover types.

The prognostic value of cardiac dysfunction assessed by bedside echocardiography in critically ill patients with COPD requiring mechanical ventilation: a study protocol

IntroductionChronic obstructive lung disease is not only a major cause of morbidity and mortality, but is also the major reason for intensive care unit (ICU) admission. Cardiac function is often...

Incidental imaging findings from routine chest CT used to identify subjects at high risk of future cardiovascular events.

To investigate the contribution of incidental findings at chest computed tomography (CT) in the detection of subjects at high risk for cardiovascular disease (CVD) by deriving and validating a CT-based prediction rule. This retrospective study was approved by the ethical review board of the primary participating facility, and informed consent was waived. The derivation cohort comprised 10 410 patients who underwent diagnostic chest CT for noncardiovascular indications. During a mean follow-up of 3.7 years (maximum, 7.0 years), 1148 CVD events (cases) were identified. By using a case-cohort approach, CT scans from the cases and from an approximately 10% random sample of the baseline cohort (n = 1366) were graded visually for several cardiovascular findings. Multivariable Cox proportional hazards analysis with backward elimination technique was used to derive the best-fitting parsimonious prediction model. External validation (discrimination, calibration, and risk stratification) was performed in a separate validation cohort (n = 1653). The final model included patient age and sex, CT indication, left anterior descending coronary artery calcifications, mitral valve calcifications, descending aorta calcifications, and cardiac diameter. The model demonstrated good discriminative value, with a C statistic of 0.71 (95% confidence interval: 0.68, 0.74) and a good overall calibration, as assessed in the validation cohort. This imaging-based model allows accurate stratification of individuals into clinically relevant risk categories. Structured reporting of incidental CT findings can mediate accurate stratification of individuals into clinically relevant risk categories and subsequently allow those at higher risk of future CVD events to be distinguished.

Sensitivity Analysis of Flux Cored arc Welding Process Variables in Super Duplex Stainless Steel Claddings

Weld surfacing with super duplex grade stainless steel found to improve corrosion resistance and functional life of the mild steel components used in the process industries. The properties of the deposited layer were influenced by the process variables that affect the heat input to the process. The influence exerted by the process variables on the responses of super duplex stainless steel claddings were modeled using the response surface models. The response surface models developed by the regression techniques using the data collected from central composite rotatable design of experiments. The data extracted from 32 single bead on the plate welds were deposited by flux cored arc welding process. The developed models can be used to predict and simulate the influence of the process variables on the responses. The insignificant variables found in the full models were removed by the backward elimination technique. Sensitivity analysis performed on the reduced models helps to identify and rank the process variables based on their extent of influence on the responses. Then the ranked variables are closely regulated to tailor the properties of the surfaced layer.

Do cell phones, iPods/MP3 players, siblings and friends matter? Predictors of child body mass in a U.S. Southern Border City Middle School

This study examines the association of children's (i) micro-social environment, specifically siblings [kin-friends] and friends from school and neighborhood [non-kin-friends], and (ii) ownership of information and communication technologies (ICT), specifically cell phones and iPod/MP3 players, with body mass index percentile (BMIp). Fifty-five randomly selected 6th graders with a mean age of 12 years, stratified by gender (23 boys and 32 girls), from a Texas middle school located in a city along the U.S. southern border. The linear regression of BMIp on number of siblings and of non-kin-friends, and ownership of cell phone and of iPod/MP3 player was examined using two models: M1 was based on the manual selection of predictors from a pool of potential predictors. M2 was derived from the predictors specified in M1 using backward elimination technique. Because sample size was small, the significance of regression coefficients was evaluated using robust standard errors to calculate t-values. Data for predictors were obtained through a survey. Height and weight were obtained through actual anthropometric measurements. BMIp was calculated using the on-line BMI calculator of the Center for Disease Control and Prevention. Findings reveal that children's social environment and ICT ownership predict BMIp; specifically, number of siblings (M2: β = -0.34, p-value < .001), and ownership of iPod/MP3 players (M2: β = 0.33, p-value < .001). These results underscore the importance of family in configuring, and of new personal technical devices (that encourage solitary, and oftentimes sedentary, activities) in predicting child body mass.

The use of feature selection based data mining methods in biomarkers identification of disease

Feature selection based data mining methods is one of the most important research directions in the fields of machine learning especially in recent years. We found that feature selection based data mining methods better suit to identifying biomarkers for disease as well as syndrome in Traditional Chinese Medicine. In this paper, we presented a novel computational strategy to select biomarkers as few as possible for disease. Firstly, we compared the three types of feature selection based data mining methods, i.e., Filter, Wrapper and Embedded methods and using 3 fold cross validation to evaluate computational performances. Alternatively, we combined independent t test and classification based data mining methods as well as backward elimination technique to select as few as possible biomarkers with best classification performances. By the novel method, we select least biomarkers for disease. And found the associated biomedical literatures support the finding. The novel method presented here provides a better insight into the pathology of a disease.

Response Surface Based Optimization Approach for Thermal Placement Design of Chips in Multiple-Chip Modules

This paper attempts to perform thermal enhancement of planar multiple-chip modules (MCMs) containing a number of chips of equal and/or unequal power through optimal chip placement design. To achieve the goal, an effective design approach is presented for the thermal design optimization problems in the context of models of placement of chips in MCMs. The approach combines the use of the currently proposed response surface (RS) based methodology, which is an optimization algorithm and a finite element modeling technique. The proposed RS-based methodology is used for creating a macro mathematical expression of the design objective of the thermal optimization problem, i.e., the total chip junction temperature of the system, associated with the design parameters, including the chip location and power. The validity of the mathematical expressions constructed is verified through two approaches. Furthermore, to make the constructed mathematical expression more compact while maintaining the associated solution accuracy, the backward variable elimination technique is employed. The effectiveness of the proposed design optimization methodology is demonstrated through several design case studies involving planar plastic ball grid array type MCMs. It is found that the proposed RS-based methodology could accurately define the macro mathematical model of the total system chip junction temperature in terms of the chip location and power. In addition, results show that the current optimal chip placement design can provide a minimal system temperature.

Predicting Cholesterol, Progesterone, and Days to Ovulation Using Postpartum Metabolic and Endocrine Measures

The objective of this study was to examine relationships among metabolic and endocrine factors that may influence ovarian activity during early lactation. Holstein cows (n=19) were bled twice each week to determine plasma concentrations of insulin (INS), glucose, cholesterol, insulin-like growth factor-1 (IGF-I), and progesterone (P4). Feed intake and milk production were recorded daily while body weights and milk composition were recorded weekly. Relationships among plasma cholesterol and P4, and days to first and second postpartum ovulation were modeled with energy balance (EB), dry matter intake, milk yield and composition, plasma metabolites, and hormones using the backward elimination technique of multivariate regression analysis. Variables that contributed the most to predicting plasma cholesterol concentrations were dry matter intake×SNF using model 1 (production variables) and the square of glucose (i.e., glucose2) using model 3 (plasma hormones and metabolites). For plasma P4 concentrations, EB (model 2, production variables) and IGF-I (model 3, plasma hormones and metabolites) were the major predictors. The production variables EB and percentage of milk lactose were the greatest contributors to the models predicting days to first and second postpartum ovulations, respectively. Of the plasma hormones and metabolites evaluated, IGF-I2 was the most significant predictor of days to first postpartum ovulation, whereas glucose2 and INS were the significant predictors of days to second postpartum ovulation. Plasma IGF-I, glucose, and INS have been implicated in ovarian functions and their significant contributions to these models are consistent with possible important roles in postpartum return to ovarian competence.

Monthly mean pressure reconstruction for the Late Maunder Minimum Period (AD 1675-1715)

The Late Maunder Minimum (LMM; 1675–1715) delineates a period with marked climate variability within the Little Ice Age in Europe. Gridded monthly mean surface pressure fields were reconstructed for this period for the eastern North Atlantic–European region (25°W–30°E and 35–70°N). These were based on continuous information drawn from proxy and instrumental data taken from several European data sites. The data include indexed temperature and rainfall values, sea ice conditions from northern Iceland and the Western Baltic. In addition, limited instrumental data, such as air temperature from central England (CET) and Paris, reduced mean sea level pressure (SLP) at Paris, and monthly mean wind direction in the Øresund (Denmark) are used. The reconstructions are based on a canonical correlation analysis (CCA), with the standardized station data as predictors and the SLP pressure fields as predictand. The CCA-based model was performed using data from the twentieth century. The period 1901–1960 was used to calibrate the statistical model, and the remaining 30 years (1961–1990) for the validation of the reconstructed monthly pressure fields. Assuming stationarity of the statistical relationships, the calibrated CCA model was then used to predict the monthly LMM SLP fields. The verification results illustrated that the regression equations developed for the majority of grid points contain good predictive skill. Nevertheless, there are seasonal and geographical limitations for which valid spatial SLP patterns can be reconstructed. Backward elimination techniques indicated that Paris station air pressure and temperature, CET, and the wind direction in the Øresund are the most important predictors, together sharing more than 65% of the total variance. The reconstructions are compared with additional data and subjectively reconstructed monthly pressure charts for the years 1675–1704. It is shown that there are differences between the two approaches. However, for extreme years the reconstructions are in good agreement. Copyright © 2000 Royal Meteorological Society

Do simple prudent health behaviours protect men from myocardial infarction?

We tested whether behaviours such as discarding obvious fat on meat, cessation of smoking, avoidance of passive smoking, habitual use of reduced fat milk, prudent consumption of alcohol and regular but moderate physical exercise are associated with a reduction of cardiovascular risk. This was a population-based case-control study done in Perth, Western Australia. The cases (n = 336) were men aged 27-64 years with a first-ever acute myocardial infarction (AMI) during the period 1992-1993, and who survived at least 28 days. The controls (n = 735) were participants in a population-based survey of cardiovascular risk factors conducted during May-November 1994. Both groups completed the same questionnaire and the data were analysed with multiple logistic regression using backward elimination technique. Among men aged 27-64 years simple measures such as participation in nonvigorous exercise (odds ratio [OR] = 0.5, 95% CI: 0.4-0.7), and avoidance of added salt (OR = 0.6, 95% CI: 0.4-0.9) are associated with significant and important protection from AMI. After 25 years of falling mortality in Australia, lifestyles can still be significantly improved to reduce heart disease even further.