• Journal of Intelligence and Information Systems
• /
• v.24 no.2
• /
• pp.111-124
• /
• 2018

This study aims to develop a classification model for predicting the occurrence of hyperlipidemia, one of the chronic diseases. Prior studies applying data mining techniques for predicting disease can be classified into a model design study for predicting cardiovascular disease and a study comparing disease prediction research results. In the case of foreign literatures, studies predicting cardiovascular disease were predominant in predicting disease using data mining techniques. Although domestic studies were not much different from those of foreign countries, studies focusing on hypertension and diabetes were mainly conducted. Since hypertension and diabetes as well as chronic diseases, hyperlipidemia, are also of high importance, this study selected hyperlipidemia as the disease to be analyzed. We also developed a model for predicting hyperlipidemia using SVM and meta learning algorithms, which are already known to have excellent predictive power. In order to achieve the purpose of this study, we used data set from Korea Health Panel 2012. The Korean Health Panel produces basic data on the level of health expenditure, health level and health behavior, and has conducted an annual survey since 2008. In this study, 1,088 patients with hyperlipidemia were randomly selected from the hospitalized, outpatient, emergency, and chronic disease data of the Korean Health Panel in 2012, and 1,088 nonpatients were also randomly extracted. A total of 2,176 people were selected for the study. Three methods were used to select input variables for predicting hyperlipidemia. First, stepwise method was performed using logistic regression. Among the 17 variables, the categorical variables(except for length of smoking) are expressed as dummy variables, which are assumed to be separate variables on the basis of the reference group, and these variables were analyzed. Six variables (age, BMI, education level, marital status, smoking status, gender) excluding income level and smoking period were selected based on significance level 0.1. Second, C4.5 as a decision tree algorithm is used. The significant input variables were age, smoking status, and education level. Finally, C4.5 as a decision tree algorithm is used. In SVM, the input variables selected by genetic algorithms consisted of 6 variables such as age, marital status, education level, economic activity, smoking period, and physical activity status, and the input variables selected by genetic algorithms in artificial neural network consist of 3 variables such as age, marital status, and education level. Based on the selected parameters, we compared SVM, meta learning algorithm and other prediction models for hyperlipidemia patients, and compared the classification performances using TP rate and precision. The main results of the analysis are as follows. First, the accuracy of the SVM was 88.4% and the accuracy of the artificial neural network was 86.7%. Second, the accuracy of classification models using the selected input variables through stepwise method was slightly higher than that of classification models using the whole variables. Third, the precision of artificial neural network was higher than that of SVM when only three variables as input variables were selected by decision trees. As a result of classification models based on the input variables selected through the genetic algorithm, classification accuracy of SVM was 88.5% and that of artificial neural network was 87.9%. Finally, this study indicated that stacking as the meta learning algorithm proposed in this study, has the best performance when it uses the predicted outputs of SVM and MLP as input variables of SVM, which is a meta classifier. The purpose of this study was to predict hyperlipidemia, one of the representative chronic diseases. To do this, we used SVM and meta-learning algorithms, which is known to have high accuracy. As a result, the accuracy of classification of hyperlipidemia in the stacking as a meta learner was higher than other meta-learning algorithms. However, the predictive performance of the meta-learning algorithm proposed in this study is the same as that of SVM with the best performance (88.6%) among the single models. The limitations of this study are as follows. First, various variable selection methods were tried, but most variables used in the study were categorical dummy variables. In the case with a large number of categorical variables, the results may be different if continuous variables are used because the model can be better suited to categorical variables such as decision trees than general models such as neural networks. Despite these limitations, this study has significance in predicting hyperlipidemia with hybrid models such as met learning algorithms which have not been studied previously. It can be said that the result of improving the model accuracy by applying various variable selection techniques is meaningful. In addition, it is expected that our proposed model will be effective for the prevention and management of hyperlipidemia.

• Tuberculosis and Respiratory Diseases
• /
• v.43 no.6
• /
• pp.882-893
• /
• 1996

Background : Over one-third of solitary pulmonary nodules are malignant, but most malignant SPNs are in the early stages at diagnosis and can be cured by surgical removal. Therefore, early diagnosis of malignant SPN is essential for the lifesaving of the patient. The incidence of pulmonary tuberculosis in Korea is somewhat higher than those of other countries and a large number of SPNs are found to be tuberculoma. Most primary physicians tend to regard newly detected solitary pulmonary nodule as tuberculoma with only noninvasive imaging such as CT and they prefer clinical observation if the findings suggest benignancy without further invasive procedures. Many kinds of noninvasive procedures for confirmatory diagnosis have been introduced to differentiate malignant SPNs from benign ones, but none of them has been satisfactory. FOG-PET is a unique tool for imaging and quantifying the status of glucose metabolism. On the basis that glucose metabolism is increased in the malignant transfomled cells compared with normal cells, FDG-PET is considered to be the satisfactory noninvasive procedure which can differentiate malignant SPNs from benign SPNs. So we performed FOG-PET in patients with solitary pulmonary nodule and evaluated the diagnostic accuracy in the diagnosis of malignant SPNs. Method : 34 patients with a solitary pulmonary nodule less than 6 cm of irs diameter who visited Samsung Medical Center from Semptember, 1994 to Semptember, 1995 were evaluated prospectively. Simple chest roentgenography, chest computer tomography, FOG-PET scan were performed for all patients. The results of FOG-PET were evaluated comparing with the results of final diagnosis confirmed by sputum study, PCNA, fiberoptic bronchoscopy, or thoracotomy. Results : (I) There was no significant difference in nodule size between malignant (3.1 1.5cm) and benign nodule(2.81.0cm)(p>0.05). (2) Peal SUV(standardized uptake value) of malignant nodules (6.93.7) was significantly higher than peak SUV of benign nodules(2.71.7) and time-activity curves showed continuous increase in malignant nodules. (3) Three false negative cases were found among eighteen malignant nodule by the FDG-PET imaging study and all three cases were nonmucinous bronchioloalveolar carcinoma less than 2 em diameter. (4) FOG-PET imaging resulted in 83% sensitivity, 100% specificity, 100% positive predictive value and 84% negative predictive value. Conclusion: FOG-PET imaging is a new noninvasive diagnostic method of solitary pulmonary nodule thai has a high accuracy of differential diagnosis between malignant and benign nodule. FDG-PET imaging could be used for the differential diagnosis of SPN which is not properly diagnosed with conventional methods before thoracotomy. Considering the high accuracy of FDG-PET imaging, this procedure may play an important role in making the dicision to perform thoracotomy in diffcult cases.

• Journal of Distribution Science
• /
• v.12 no.4
• /
• pp.31-39
• /
• 2014

Purpose - This study investigates the following two issues. First, we attempt to find the important determinants of housing investment and to identify their significance rank using survey panel data. Recently, the expansion of global uncertainty in the real estate market has directly and indirectly influenced the Korean housing market; households demonstrate a sensitive reaction to changes in that market. Therefore, this study aims to draw conclusions from understanding how the impact of financial strength of the household is related to house investment. Second, we attempt to verify the effectiveness of diverse indices of financial strength such as DTI, LTV, and PIR as measures to monitor the housing market. In the continuous housing market recession after the global crisis, the government places top priority on residence stability. However, the government still imposes forceful restraints on indices of financial strength. We believe this study verifies the utility of these regulations when used in the housing market. Research design, data, and methodology - The data source for this study is the "National Survey of Tax and Benefit" from 2007 (1st) to 2011 (5th) by the Korea Institute of Public Finance. Based on this survey data, we use panel data of 3,838 households that have been surveyed continuously for 5 years. We sort the base variables according to relevance of house investment criteria using the decision tree model (DTM), which is the standard decision-making model for data-mining techniques. The DTM method is known as a powerful methodology to identify contributory variables for predictive power. In addition, we analyze how important explanatory variables and the financial strength index of households affect housing investment with the binary logistic multi-regressive model. Based on the analyses, we conclude that the financial strength index has a significant role in house investment demand. Results - The results of this research are as follows: 1) The determinants of housing investment are age, consumption expenditures, income, total assets, rent deposit, housing price, habits satisfaction, housing scale, number of household members, and debt related to housing. 2) The impact power of these determinants has changed more or less annually due to economic situations and housing market conditions. The level of consumption expenditure and income are the main determinants before 2009; however, the determinants of housing investment changed to indices of the financial strength of households, i.e., DTI, LTV, and PIR, after 2009. 3) Most of all, since 2009, housing loans has been a more important variable than the level of consumption in making housing market decisions. Conclusions - The results of this research show that sound financing of households has a stronger effect on housing investment than reduced consumption expenditures. At the same time, the key indices that must be monitored by the government under economic emergency conditions differ from those requiring monitoring under normal market conditions; therefore, political indices to encourage and promote the housing market must be divided based on market conditions.

• Journal of Korean Neurosurgical Society
• /
• v.63 no.1
• /
• pp.89-98
• /
• 2020

Objective : Ossification of posterior longitudinal ligament (OPLL) in the thoracic spine may cause chronic compressive myelopathy that is usually progressive, and unfavorable by conservative treatment. Although surgical intervention is often needed, the standard surgical method has not been established. Recently, it has been reported that posterior decompression with dekyphosis is effective surgical technique for favorable clinical outcome. The purpose of this study was to evaluate the surgical outcomes in patients with thoracic OPLL according to dekyphosis procedure and to identify predictive factors for the surgical results. Methods : A total of 25 patients with thoracic OPLL who underwent surgery for myelopathy from May 2004 to March 2017, were retrospectively reviewed. Patients with cervical myelopathy were excluded. We assessed the clinical outcomes according to various surgical approaches. The modified Japanese orthopedic association (JOA) scores for the thoracic spine (total, 11 points) and JOA recovery rates were used for investigating surgical outcomes. Results : Of the 25 patients, 10 patients were male and the others were female. The mean JOA score was 6.7±2.3 points preoperatively and 8.8±1.8 points postoperatively, yielding a mean recovery rate of 53.8±31.0%. The mean patients' age at surgery was 52.4 years and mean follow-up period was 40.2 months. According to surgical approaches, seven patients underwent anterior approaches, 13 patients underwent posterior approaches, five patients underwent combined approaches. There was no significant difference of the surgical outcomes related with different surgical approaches. Age (≥55 years) and high signal intensity on preoperative magnetic resonance (MR) image in the thoracic spine were significant predictors of the lower recovery rate after surgery (p<0.05). Posterior decompression with dekyphosis procedure was related to the excellent surgical outcomes (p=0.047). Dekyphosis did not affect the complication rates. Conclusion : In this study, our result elucidated that old age (≥55 years) and presence of intramedullary high signal intensity on preoperative MR images were risk factors related to poor surgical outcomes. In the meanwhile, posterior decompression with dekyphosis affected favorable clinical outcome. Posterior approach with dekyphosis procedure can be a recommendable surgical option for favorable results.

• Asian Pacific Journal of Cancer Prevention
• /
• v.17 no.10
• /
• pp.4623-4629
• /
• 2016

Introduction: Breast cancer is multifactorial therefore more recognition of risk factors is important in its prevention. Objective: This study was conducted in order to determine the factors influencing breast cancer in women referred to health centers in Guilan province in 2015-2016. Method: In a case- control study, 225 women with breast cancer were investigated. The control group consisted of 225 healthy women of the relatives (third-rank) whose phone numbers were obtained from the patients. Data were collected through telephone interviews. Results: The risk of breast cancer raised in women who have a family history of other cancers (OR= 3.5; 95% CI= 1.96-6.6), exposure to X-Ray (OR= 2.5; 95% CI=1.1-5.5), having more than 4 children (OR= 2.695% CI=1.2-4.8), age more than 36 years at first pregnancy(OR=2.3; 95% CI=0.7-5.1),primary levelof education (OR= 5.4;95% CI=2.8-11.2) and inadequate intake of fruit (OR=1.5; 95% CI=1-2.2). Also, presence of the following factors reduced breast cancer risk: regular menstruation (OR= 0.66; CI=0.4-0.9), duration of breastfeeding more than 12 months, less than 6 months and 7-12 months (OR=0.23; 95% CI=0.09-0.59, OR=0.29; 95% CI=0.17-0.49 and OR=0.03; 95% CI=0.01-0.08) and parity (OR=0.4; 95% CI=0.27-0.83) In multiple linear regression analysis of higher education (OR=0.16; 95% CI=0.03-0.77), using contraceptives for more than 16 years (OR=2.3; 95% CI=1.4-3.9), family history of other cancers (OR=6.1; 95% CI=1.9-19.3) and a history of X-Ray exposure (OR=4.4; 95% CI=1.07-18.1) were considered as predictive factors. Conclusion: The results of this study emphasize the importance of informing women about breast cancer risk factors. So, identification of these risk factors is required as important means of prevention and treatment of breast cancer.

• Maxillofacial Plastic and Reconstructive Surgery
• /
• v.33 no.3
• /
• pp.233-240
• /
• 2011

Purpose: To apply a computer assisted navigation system to orthognathic surgery, a simple and efficient measuring algorithm calculation based on affine transformation was designed. A method of improving accuracy and reducing errors in orthognathic surgery by use of an optical tracking camera was studied. Methods: A total of 5 points on one surgical splint were measured and tracked by the Polaris $Vicra^{(R)}$ (Northern Digital Inc Co., Ontario, Canada) optical tracking system in two cases. The first case was to apply the transformation matrix at pre- and postoperative situations, and the second case was to apply an affine transformation only after the postoperative situation. In each situation, the predictive measuring value was changed to the final measuring value via an affine transformation algorithm and the expected coordinates calculated from the model were compared with those of the patient in the operation room. Results: The mean measuring error was $1.027{\pm}0.587$ using the affine transformation at pre- and postoperative situations and the average value after the postoperative situation was $0.928{\pm}0.549$. The farther a coordinate region was from the reference coordinates which constitutes the transform matrixes, the bigger the measuring error was found which was calculated from an affine transformation algorithm. Conclusion: Most difference errors were brought from mainly measuring process and lack of reproducibility, the affine transformation algorithm formula from postoperative measuring values by using of optic tracking system between those of model surgery and those of patient surgery can be selected as minimizing the difference error. To reduce coordinate calculation errors, minimum transformation matrices must be used and reference points which determine an affine transformation must be close to the area where coordinates are measured and calculated, as well as the reference points need to be scattered.

• Tuberculosis and Respiratory Diseases
• /
• v.39 no.2
• /
• pp.150-158
• /
• 1992

Background: Although many authors have reported that the median survival time of surgically resected non-small cell lung cancer (NSCLC) was shorter in aneuploid than in diploid determined by flow cytometry, there are few reports about DNA ploidy using bronchial brushing material in all types of lung cancer. Method: The DNA ploidy test results of 109 consecutive patients with lung cancer were analyzed to find the relationship of DNA ploidy and anatomic or physiologic stage. And the differences of the response to various therapeutic modalities according to DNA ploidy were evaluated at least 8 weeks after the begining of the therapy. Results: Numbers of patients with DNA aneuploid pattern or high proliferative activity (S+G2M>22%) were not different among the various cell types of lung cancer. The relationship of DNA ploidy and anatomic or physiologic stage was not significant. However, NSCLC patients with high proliferative activity showed more advanced anatomic stage than those without that (p<0.05). The short-term response rate to therapy depended on the anatomic (p<0.005) or physiologic stages (p<0.05) in patients with NSCLC, and not on DNA ploidy or proliferative activity. Conclusion: DNA ploidy test using bronchial brushing material revealed that high proliferative activity means advanced anatomic stage, but it was not useful to predict the therapeutic response.

• Clinics in Shoulder and Elbow
• /
• v.5 no.2
• /
• pp.88-97
• /
• 2002

The purpose of this study was to compare the outcomes between arthroscopir repair and mini-open repair of medium and large rotator cuff tears in which arthroscopic repair was technically unsuccessful. We evaluated 76 patients of full-thickness rotator cuff tears, among them 42 patients had all-arthroscopic and 34 patients had mini-open salvage repairs. Patients who had acromioclavicular arthritis, subscapularis tear, or instability were excluded. There were 39 males and 37 females with mean age of 56 years (range,42 to 75 years). At a mean follow-up of 39 months (range, 24 to 64 months), the results of both groups were compared with regard to the UCLA and ASES shoulder rating scale s. Shoulder scores improved in all ratings in both groups (p > 0.05). Overall, sixty-six patients showed excellent or gr)of and ten patients showed fair or poor scores by the UCLA scale. Seventy-two patients satisfactorily returned to prior activity. Four showed unsatisfactory return. The range of motion, strength, and patient's satisfaction were improved postoperatively. There were no difference in shoulder scores, pain, and activity return between the arthroscopic and mini-open salvage groups (p > 0.05). However, Patients with larger size tear showed lower shoulder scores and less predictive recovery of the strength and function (p < 0.05). Postoperative pain was not different with respect to the size of the tear (p : 0.251). Arthroscopic repair of medium and large full-thickness rotator cuff tears had iln equal outcome to technically unsuccessful arthroscopic repairs, which were salvaged by conversion to a mini- open repair technique. Surgical outcome depended on the size of the tear, rather than the method of repair.

• Journal of Crop Science and Biotechnology
• /
• v.11 no.2
• /
• pp.141-150
• /
• 2008

The response of grain yield(GY) and milled-rice protein content(PC) to crop growth status and nitrogen(N) rates at panicle initiation stage(PIS) is critical information for prescribing topdress N rate at PIS(Npi) for target GY and PC. Three split-split-plot experiments including various N treatments and rice cultivars were conducted in Experimental Farm, Seoul National University, Korea in 2003-2005. Shoot N density(SND, g N in shoot $m^{-2}$) and canopy reflectance were measured before N application at PIS, and GY, PC, and SND were measured at harvest. Data from the first two years(2003-2004) were used for calibrating the predictive models for GY, PC, and SND accumulated from PIS to harvest using SND at PIS and Npi by multiple stepwise regression. After that the calibrated models were used for calculating N requirement at PIS for each of nine plots based on the target PC of 6.8% and the values of SND at PIS that was estimated by canopy reflectance method in the 2005 experiment. The result showed that SND at PIS in combination with Npi were successful to predict GY, PC, and SND from PIS to harvest in the calibration dataset with the coefficients of determination ($R^2$) of 0.87, 0.73, and 0.82 and the relative errors in prediction(REP, %) of 5.5, 4.3, and 21.1%, respectively. In general, the calibrated model equations showed a little lower performance in calculating GY, PC, and SND in the validation dataset(data from 2005) but REP ranging from 3.3% for PC and 13.9% for SND accumulated from PIS to harvest was acceptable. Nitrogen rate prescription treatment(PRT) for the target PC of 6.8% reduced the coefficient of variation in PC from 4.6% in the fixed rate treatment(FRT, 3.6g N $m^{-2}$) to 2.4% in PRT and the average PC of PRT was 6.78%, being very close to the target PC of 6.8%. In addition, PRT increased GY by 42.1 $gm^{-2}$ while Npi increased by 0.63 $gm^{-2}$ compared to the FRT, resulting in high agronomic N-use efficiency of 68.8 kg grain from additional kg N. The high agronomic N-use efficiency might have resulted from the higher response of grain yield to the applied N in the prescribed N rate treatment because N rate was prescribed based on the crop growth and N status of each plot.

• The Korean Journal of Applied Statistics
• /
• v.21 no.1
• /
• pp.141-149
• /
• 2008

Recently we frequently analyze multivariate data with quite large number of variables. In such data sets, virtually duplicated variables may exist simultaneously even though they are conceptually distinguishable. Duplicate variables may cause problems such as the distortion of principal axes in principal component analysis and factor analysis and the distortion of the distances between observations, i.e. the input for cluster analysis. Also in supervised learning or regression analysis, duplicated explanatory variables often cause the instability of fitted models. Since real data analyses are aimed often at multiple purposes, it is necessary to reduce the number of variables to a parsimonious level. The aim of this paper is to propose a practical algorithm for selection of a subset of variables from a given set of p input variables, by the criterion of minimum trace of partial variances of unselected variables unexplained by selected variables. The usefulness of proposed method is demonstrated in visualizing the relationship between selected and unselected variables, in building a predictive model with very large number of independent variables, and in reducing the number of variables and purging/merging categories in categorical data.