• Journal of Intelligence and Information Systems
• /
• v.23 no.1
• /
• pp.127-141
• /
• 2017

Advances in Internet technologies and the proliferation of mobile devices enabled consumers to approach a wide range of goods and services, while causing an adverse effect that they have hard time reaching their congenial items even if they devote much time to searching for them. Accordingly, businesses are using the recommender systems to provide tools for consumers to find the desired items more easily. Association Rule Mining (ARM) technology is advantageous to recommender systems in that ARM provides intuitive form of a rule with interestingness measures (support, confidence, and lift) describing the relationship between items. Given an item, its relevant items can be distinguished with the help of the measures that show the strength of relationship between items. Based on the strength, the most pertinent items can be chosen among other items and exposed to a given item's web page. However, the diversity of the measures may confuse which items are more recommendable. Given two rules, for example, one rule's support and confidence may not be concurrently superior to the other rule's. Such discrepancy of the measures in distinguishing one rule's superiority from other rules may cause difficulty in selecting proper items for recommendation. In addition, in an online environment where a web page or mobile screen can provide a limited number of recommendations that attract consumer interest, the prudent selection of items to be included in the list of recommendations is very important. The exposure of items of little interest may lead consumers to ignore the recommendations. Then, such consumers will possibly not pay attention to other forms of marketing activities. Therefore, the measures should be aligned with the probability of consumer's acceptance of recommendations. For this reason, this study proposes a model-based approach to combine those measures into one unified measure that can consistently determine the ranking of recommended items. A regression model was designed to describe how well the measures (independent variables; i.e., support, confidence, and lift) explain consumer's acceptance of recommendations (dependent variables, hit rate of recommended items). The model is intuitive to understand and easy to use in that the equation consists of the commonly used measures for ARM and can be used in the estimation of hit rates. The experiment using transaction data from one of the Korea's largest online shopping malls was conducted to show that the proposed model can improve the hit rates of recommendations. From the top of the list to 13th place, recommended items in the higher rakings from the proposed model show the higher hit rates than those from the competitive model's. The result shows that the proposed model's performance is superior to the competitive model's in online recommendation environment. In a web page, consumers are provided around ten recommendations with which the proposed model outperforms. Moreover, a mobile device cannot expose many items simultaneously due to its limited screen size. Therefore, the result shows that the newly devised recommendation technique is suitable for the mobile recommender systems. While this study has been conducted to cover the cross-selling in online shopping malls that handle merchandise, the proposed method can be expected to be applied in various situations under which association rules apply. For example, this model can be applied to medical diagnostic systems that predict candidate diseases from a patient's symptoms. To increase the efficiency of the model, additional variables will need to be considered for the elaboration of the model in future studies. For example, price can be a good candidate for an explanatory variable because it has a major impact on consumer purchase decisions. If the prices of recommended items are much higher than the items in which a consumer is interested, the consumer may hesitate to accept the recommendations.

• Journal of Korea Water Resources Association
• /
• v.42 no.6
• /
• pp.457-464
• /
• 2009

This paper proposes methodologies for analyzing the accuracy of the proportional hazards model in predicting consecutive break times of water mains and estimating the time interval for economical water main replacement. By using the survival functions that are based on the proportional hazards models a criterion for the prediction of the consecutive pipe breaks is determined so that the prediction errors are minimized. The criterion to predict pipe break times are determined as the survival probability of 0.70 and only the models for the third through the seventh break are analyzed to be reliable for predicting break times for the case study pipes. Subsequently, the criterion and the estimated lower and upper bound survival functions of consecutive breaks are used in predicting the lower and upper bounds of the 95% confidence interval of future break times of an example water main. Two General Pipe Break Prediction Models(GPBMs) are estimated for an example pipe using the two series of recorded and predicted lower and upper bound break times. The threshold break rate is coupled with the two GPBMs and solved for time to obtain the economical replacement time interval.

• Korean Journal of Clinical Pharmacy
• /
• v.22 no.4
• /
• pp.340-346
• /
• 2012

Background: Amphotericin B is a mainstay in the treatment of many systemic fungal infections due to its wide antifungal spectrum and low incidence of resistance. However, the use of amphotericin B is limited by its nephrotoxicity. Objectives: The objective of this study was to evaluate the incidence and risk factors of renal adverse drug reactions (ADRs) of conventional amphotericin B (Fungizone$^{(R)}$). In addition, we compared the changes of serum creatinine (SCr) between patients who remained conventional amphotericin B and patients who were switched to liposomal amphotericin B after occurrence of renal adverse reactions. Methods: Adult hospitalized patients who reported renal adverse reactions caused by conventional amphotericin B from January 2011 to July 2012 at pharmacovigilance center in Yonsei University Healthcare System included in this study. ADRs scored as 'doubtful' in Naranjo probability ADR scale were excluded. We retrospectively analyzed patients' basic clinical characteristics, concurrent diseases or nephrotoxic drugs in order to find variables that can correlate with occurrence of renal ADRs. Changes in SCr were compared between conventional amphotericin B group and liposomal amphotericin B group. Results: A total of 231 ADRs after administration of conventional amphotericin B in 75 patients were reported to pharmacovigilance center and assessed their severities as 'possible', 'probable', or 'definite'. Renal adverse reaction was the most common ADR with incidence rate of 42% (96 of 231 ADRs). Mean change in SCr from baseline was 0.26 mg/dL (change % 37.8) and statistically significant (p=0.000). Simple correlations analysis revealed that the number of concurrent diseases and number of nephrotoxic drugs were positively correlated with changes in SCr, but these results were not statistically significant. Among 43 patients who remained amphotericin B after occurrence of renal ADRs, 27 patients was administered conventional amphotericin B and 16 patients changed to liposomal amphotericin B. Mean change in SCr in amphotericin B group was 0.23 mg/dL (32.75%), whereas mean change in SCr in liposomal amphotericin B group were -0.28 mg/dL (19.38%) and difference between two groups was statistically significant (p=0.003). The numbers of patient with SCr elevation more than 30% were 9 (33.3%) in amphotericin B group and 2 (12.5%) in liposomal amphotericin B group (Odd Ratio=3.50, 95% Confidence Interval 0.65-18.85; p=0.130). Conclusion: An analysis of ADRs due to amphotericin B administration revealed significant mean changes in SCr from baseline. Switching to liposomal amphotericin B showed significant decrease in SCr compared with conventional amphotericin B.

• Journal of Chest Surgery
• /
• v.53 no.3
• /
• pp.104-113
• /
• 2020

Background: Accurate intraoperative assessment of mediastinal lymph nodes is a critical aspect of lung cancer surgery. The efficacy and potential for upstaging implicit in these dissections must therefore be revisited in the current era of uniportal video-assisted thoracoscopic surgery (VATS). Methods: A retrospective study was conducted in which 544 patients with stage I (T1abc-T2a, N0, M0) primary lung cancer were analyzed. To assess risk factors for nodal upstaging and to limit any imbalance imposed by surgical choices, we constructed an inverse probability of treatment-weighted (IPTW) logistic regression model (in addition to non-weighted logistic models). We also evaluated risk factors for early locoregional recurrence using IPTW logistic regression analysis. Results: In the comparison of uniportal and multiportal VATS, the resected lymph node count (14.03±8.02 vs. 14.41±7.41, respectively; p=0.48) and rate of nodal upstaging (6.5% vs. 8.7%, respectively; p=0.51) appeared similar. Predictors of nodal upstaging included tumor size (odds ratio [OR], 1.74; 95% confidence interval [CI], 1.12-2.70), carcinoembryonic antigen level (OR, 1.11; 95% CI, 1.04-1.18), and histologically confirmed pleural invasion (OR, 3.97; 95% CI, 1.89-8.34). The risk factors for locoregional recurrence within 1 year were found to be number of resected N2 nodes, age, and nodal upstaging. Conclusion: Uniportal and multiportal VATS appear similar with regard to accuracy and thoroughness, showing no significant difference in the extent of nodal dissection.

• Asian Pacific Journal of Cancer Prevention
• /
• v.16 no.10
• /
• pp.4339-4345
• /
• 2015

Occupational exposure to wood dust is one cause of nasopharyngeal cancer (NPC); however, assessing this exposure remains problematic. Therefore, the objective of this study was to develop a semi-quantitative exposure assessment method and then utilize it to evaluate the association between occupational exposure to wood dust and the development of NPC. In addition, variations in risk by histology were examined. A case-control study was conducted with 327 newly diagnosed cases of NPC at the National Cancer Institute and regional cancer centers in Thailand with 1:1 controls matched for age, gender and geographical residence. Occupational information was obtained through personal interviews. The potential probability, frequency and intensity of exposure to wood dust were assessed on a job-by-job basis by experienced experts. Analysis was performed by conditional logistic regression and presented in odds ratio (ORs) estimates and 95% confidence intervals (CI). Overall, a non significant relationship between occupational wood dust exposure and NPC risk for all subjects was observed (ORs=1.61, 95%CI 0.99-2.59); however, the risk became significant when analyses focused on types 2 and 3 of NPC (ORs=1.62, 95%CI 1.03-2.74). The significant association was stronger for those exposed to wood dust for > 10 year (ORs=2.26, 95%CI 1.10-4.63), for those with first-time exposure at age > 25 year (ORs=2.07, 95%CI 1.08-3.94), and for those who had a high cumulative exposure (ORs=2.17, 95%CI 1.03-4.58) when compared with those considered unexposed. In conclusion, wood dust is likely to be associated with an increased risk of type 2 or 3 NPC in the Thai population. The results of this study show that semi-quantitative exposure assessment is suitable for occupational exposure assessment in a case control study and complements the information from self-reporting.

• Asian-Australasian Journal of Animal Sciences
• /
• v.26 no.10
• /
• pp.1474-1483
• /
• 2013

The objective was to evaluate the use of prediction equations based on the chemical composition of feedstuffs to estimate the values of apparent metabolisable energy corrected for nitrogen balance (AMEn) of corn and soybean meal for broilers. For performance and carcass characteristics, 1,200 one-d-old birds (male and female) were allotted to a completely randomised factorial $2{\times}8$ (two genders and eight experimental diets) with three replicates of each sex with 25 birds. In the metabolism trial, 240 eight-d-old birds were distributed in the same design, but with a split plot in time (age of evaluation) with five, four and three birds per plot, respectively, in stages 8 to 21, 22 to 35, and 36 to 42 d of age. The treatments consisted of the use of six equations systems to predict the AMEn content of feedstuffs, tables of food composition and AMEn values obtained by in vivo assay, totalling eight treatments. Means were compared by Scott-Knott test at 5% probability and a confidence interval of 95% was used to check the fit of the energy values of the diets to the requirements of the birds. As a result of this study, the use of prediction equations resulted in better adjustment to the broiler requirements, resulting in better performance and carcass characteristics compared to the use of tables, however, the use of energy values of feedstuffs obtained by in vivo assay is still the most effective. The best equations were: AMEn = 4,021.8-227.55 Ash (for corn) combined with AMEn = -822.33+69.54 CP-45.26 ADF+90.81 EE (for soybean meal); AMEn = 36.21 CP+85.44 EE+37.26 NFE (nitrogen-free extract) (for corn) combined with AMEn = 37.5 CP+46.39 EE+14.9 NFE (for soybean); and AMEn = 4,164.187+51.006 EE-197.663 Ash-35.689 CF-20.593 NDF (for corn and soybean meal).

• Journal of Veterinary Clinics
• /
• v.30 no.4
• /
• pp.258-263
• /
• 2013

The primary goal of this study was to compute sample sizes required to achieve the each aim of a variety of foot-and-mouth disease (FMD) surveillance programs, using a statistically valid technique that takes the following factors into account: sensitivity (Se) and specificity (Sp) of diagnostic test system, desired minimum detectable prevalence, precision, population size, and desired power of the survey. In addition, sample sizes to detect FMD if the disease is present and also as proof of freedom were computed. The current FMD active surveillance programs consist of clinical, virological, and serological surveillance. For the 2012 serological surveillance, annual sample sizes (n = 265,065) are planned at four separate levels: statistical (n = 60,884) and targeted (n = 115,232) at breeding pig farms and slaughter house, in together with the detection of structural proteins (SP) antibodies against FMD (n = 88,949). Overall, the sample size was not designed taking the specific aims of each surveillance stream into account. The sample sizes for statistical surveillance, assuming stratified two-stage sampling technique, was based to detect at least one FMD-infected case in the general population. The resulting sample size can be used to obtain evidence of freedom from FMD infection, not for detecting animals that have antibodies against FMD virus non-structural proteins (NSP). Additionally, sample sizes for targeted surveillance were not aimed for the population at risk, and also without consideration of statistical point of view. To at least the author's knowledge, sampling plan for targeted, breeding pig farms and slaughter house is not necessary and need to be included in the part of statistical surveillance. Assuming design prevalence of 10% in an infinite population, a total of 29 animals are required to detect at least one positive with probability of 95%, using perfect diagnostic test system (Se = Sp = 100%). A total of 57,211 animals needed to be sampled to give 95% confidence of estimating SP prevalence of 80% at the individual animal-level with a precision of ${\pm}5%$, assuming 800 herds with an average 200 heads per farm, within-farm variance of 0.2, between-farm variance of 0.05, cost ratio of 100:1 of farm against animals. Furthermore, 779,736 animals were required to demonstrate FMD freedom, and the sample size can further be reduced depending on the parameters assumed.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.17 no.6
• /
• pp.130-137
• /
• 2013

This research was performed through the experimental design to get the statistical analysis on foamed concrete mixed plaster with hydrogen peroxide. In this experiment, we set the ratio of each material, which part of lightweight concrete, as experimental factors and evaluated on the mechanical properties by statistical analysis for response variables obtained from experiments. Experimental factors are plaster replacement, water binder ratio, and hydrogen peroxide ratio. Response variables are dry density, compressive strength, and flexural strength. Mixing design of the foamed concrete set up a total of 15 experimental points by Box-Behnken (BB) method of the response surface analysis. Thus, the results of a study were summarized as follows. Values of the probability in experimental factors (plaster replacement, water binder ratio and hydrogen peroxide ratio) on the response variables were estimated to be significant at the 95% of confidence limit. On response surface analysis for dry density of foamed concrete, water binder ratio and hydrogen peroxide ratio were estimated to be significant (${\alpha}$ = 0.05), and the relationship between the amount of void and the water content for dry density is inverse proportional. On response surface analysis for the compressive strength of foamed concrete, water binder ratio, hydrogen peroxide ratio and (hydrogen peroxide ratio)$^2$ was estimated to be significant (${\alpha}$ = 0.05). On response surface analysis for the flexural strength of foamed concrete, water binder ratio, hydrogen peroxide ratio was estimated to be significant (${\alpha}$ = 0.05). Through multi response surface analysis, we found the optimal area that meets performance goals.

• Journal of Navigation and Port Research
• /
• v.33 no.3
• /
• pp.181-191
• /
• 2009

In general, seafarer's human error is considered to be the preponderant muse for the majority of maritime transportation accidents in a merchant ship. The implementation techniques for Human Error Model (HEM) to assess possible accident risk by deck officers including captain, chief officer, second mate and third mate are described in this study. The scope of this work is focused to 642 deck officers in the ship management company with 130 vessels. At first, HEM can be constructed through the statistical analysis and expert's brainstorming process with human data to 642 deck officers. Then the variables $\upsilon$ for the human factors, the evaluation level EP($\upsilon$) for $\upsilon$, the weight $\alpha$ of $\upsilon$, and the title weight $\beta$ of each deck officers can be decided. In addition, through the analysis of ship's accident history, the accident causation ratios by human error ${\gamma}_H$ and by external error ${\gamma}_B$ can be found as 0.517(51.7%) and 0.483(48.3%), respectively. The correlation coefficients to $\upsilon$ are also shown significant for a 95% confidence interval (p < 0.05) for each coefficient. And the validity of HEM is also surveyed by the analysis of normal probability distribution of risk level RL to each deck officer.

• Journal of Preventive Medicine and Public Health
• /
• v.39 no.2
• /
• pp.115-122
• /
• 2006

Objectives : This study was conducted to examine the relationships of the several socioeconomic position indicators with the mortality risk in a representative longitudinal study of South Korea. Methods : The 1998 National Health and Nutrition Examination Survey was conducted on a cross-sectional probability sample of South Korean households, and it contained unique 13-digit personal identification numbers that were linked to the data on mortality from the National Statistical Office of Korea. Of 5,607 males and females, 264 died between 1999 and 2003. Cox's regression was used to estimate the relative risks (RR) and their 95% confidence intervals (CI) of mortality. Results: Socioeconomic differences in mortality were observed after adjustments were made for gender and age. Compared with those people having college or higher education, those people without any formal education had a greater mortality risk (RR=2.21, 95% CI=1.12-4.40). The mortality risk among manual workers was significantly greater than that for the non-manual workers (RR=2.73, 95% CI=1.47-5.06). A non-standard employment status was also associated with an increase in mortality: temporary or daily workers had a greater mortality risk than did the full-time workers (RR=3.01, 95% CI=1.50-6.03). The mortality risk for the low occupational class was 3.06 times greater than that of the high and middle occupational classes (95% CI=1.75-5.36). In addition, graded mortality differences according to equivalized monthly household income were found. A reduction of monthly household income by 500 thousand Korean Won (about 400 US dollars) was related with a 20% excess risk of mortality. Self-reported poor living standards were also associated with an increased risk of mortality. Those without health insurance had a 3.63 times greater risk of mortality than the insured (95% CI=1.61-8.19). Conclusions: This study showed the socioeconomic differentials in mortality in a national longitudinal study of South Korea. The existence of socioeconomic mortality inequalities requires increased social discussion on social policies in Korean society. Furthermore, the mechanisms for the socioeconomic inequalities of mortality need to be explored in future studies.