Kernel Smoothing For ROC Curve And Estimation For Thyroid Stimulating Hormone

Receiver Operating Characteristic (ROC) Curves are frequently used in biomedical informatics research to evaluate classification and prediction models to support decision, diagnosis, and prognosis. ROC analysis investigates the accuracy of models and has ability to separate positive from negative cases. It is especially useful in evaluating predictive models and compare to other tests which produce output values in a continuous range. Empirical ROC curve is jagged but a true ROC curve is smooth. For this purpose kernel smoothing were used. The Area Under ROC Curve (AUC) frequently is used as a measure of the effectiveness of diagnostic markers. In this study we compare estimation of this area based on normal assumptions and kernel smoothing. This study used measurements of TSH from patients and non-diseased people of congenital hypothyroidism screening in Isfahan province. Using the method, TSH ROC curves from Isfahani's infants were fitted. For evaluating of accuracy of this test, AUC and its standard error calculated. Also effectiveness of the kernel methods in comparison to other methods showed.

The prevalence of coronary artery disease in an urban population in Isfahan, Iran.

OBJECTIVE: Cardiovascular diseases, especially coronary artery disease (CAD), are responsible for the highest mortality rate in Iran. This study was conducted to determine the prevalence of CAD in an urban sample in Isfahan by the Minnesota code of a 12-lead resting electrocardiogram (ECG), the Rose questionnaire on chest pain and a self-reported previous medical history. METHODS AND RESULTS: Among the target sample of 6,470 men and women aged 35-79 years who were randomly selected from 80 random clusters in Isfahan, 5,773 subjects (about 90%) have participated. The WHO (Rose) questionnaires (Q) on chest pain were completed for all participants and 12-lead ECGs were taken. The overall prevalence of CAD based on the Rose Q and/or ECG was 19.4% (95% CI 18.4% to 20.4%) which was significantly higher among women 21.9% (95% CI 20.5% to 23.3%) than men 16.0% (95% CI 14.5% to 17.5%) (p < 0.05). The prevalence of CAD increased with age in both sexes. The prevalence of definite and possible angina based on the questionnaire was higher among women compared to men (p < 0.05), also a greater prevalence of ECG-based possible ischaemia was observed among woman than men (12.3% vs. 7.5%) (p < 0.05). However, definite and possible MI and definite ischaemia based on ECG abnormalities were higher among men than women (p < 0.05). The total prevalence of symptomatic CAD was 9.3% and about 22% of those with symptoms of CAD on Q have some evidence on ECG. The findings also showed that CAD is more common among people with less education, lower income and the unemployed (p < 0.05). CONCLUSION: These findings indicate that there is a high prevalence of CAD among the Iranian population which need more programmes of health promotion and lifestyle changes and further studies to assess the used epidemiological methods for estimating CAD prevalence, especially among women.

Secular trends in cardiovascular mortality in Iran, with special reference to Isfahan.

OBJECTIVE: There has been a general decline in mortality from cardiovascular diseases (CVDs) in most of the developed countries since the beginning of the 1970s. Still, in recent years developing countries have seen an increasing frequency in CVD mortality. However, mortality rate studies in these populations are scarce. Here we report all-cause and CVD mortality rates for men and women aged 25-74 years over a 16-year period in 24 cities in Iran with special reference to the city of Isfahan. METHODS AND RESULTS: The study was based on national death records using the ninth international classification of diseases and age standardization was performed using the total population of Iran in 1985 as a standard. Due to limitations in available data, mortality rates for the specific categories of CVD for the whole country could not be provided. The in-hospital death rates following myocardial infarction in coronary care units (CCUs) and cardiology departments in Isfahan hospitals were also assessed. The completed medical records from hospitals or the relatives of decedents were reviewed by physicians certified in internal medicine, cardiology and neurology to assess the reliability of death certificate data regarding CVD by determining the sensitivity and specificity of the death certificates against the standard of the reviewers. The official circulatory diseases proportional mortality ratio continues to rise since 1981 with a steep increase since 1987, constituting 26.6% and 47.3% of all deaths in 1981 and 1995, respectively. Age-adjusted all-cause and CVD mortality data were decreasing since 1981 and increasing since 1990. During those years age-adjusted CVD, stroke and other CVD mortality rates were decreasing in Isfahan with a slight increase in ischaemic heart disease (IHD) death rates in both sexes. Mortality rates based on sex showed a 38% and 24.8% decline in all-cause and CVD mortality in men between 1981 to 1995, and a 35% and 34.9% decline for female mortality rates for the same period, respectively. The in-hospital death rate following myocardial infarction in Isfahan was increasing between 1993 and 1995 with a slight decrease thereafter. The results of death certification assessment showed a specificity of 0.89 and a sensitivity of 0.43 with the positive and negative predictive values of 0.82 and 0.57, respectively. CONCLUSION: These data indicate that circulatory diseases remain a serious public health threat in Iran. It suggests the ongoing need for more regular, systematic and innovative surveillance data to improve the capability of measuring, explaining and predicting the disease trend on which the national public health policy depends.

Prevalence and risk factors of diabetes mellitus in the Isfahan city population (aged 40 or over) in 1993.

The objective of this study was to estimate the prevalence of diabetes among people aged 40 or more and to compare selected factors such as systolic and diastolic blood pressure, body mass index and lipid profile in diabetic and non-diabetic subjects. This is a cross-sectional study of subjects aged over 40 in Isfahan. Selection of diabetic subjects were based on multi-stage random sampling. Diabetes status was defined by the World Health Organization criteria and was based on fasting plasma glucose and oral glucose tolerance test results. Data on age, body mass index, blood pressure and lipid profile were obtained. Results show that a total of 3910 eligible subjects were examined and classified into different diabetes status. The overall age-standardized prevalence of diabetes was 7.54% (95% CI: 6.24-8.84%) in males and 7.97% (95% CI: 6.75-9.19%) in females with no difference. Similarly, the overall age standardized prevalence was 7.76% (95% CI: 6.87-8.65%). The prevalence of diabetes was shown to increased in the older age groups. Multiple logistic regression analysis showed that age, systolic blood pressure, hypertriglyceridemia, being overweight and also obesity were independently associated with diabetes. A large number of newly-diagnosed diabetic subjects were identified, which necessitates powerful screening programs. High levels of blood pressure, body mass index and lipid profile should draw attention to probable presence of diabetes.