• Journal of Intelligence and Information Systems
• /
• v.27 no.4
• /
• pp.73-95
• /
• 2021

This study uses Node2vec graph embedding method and Light GBM link prediction to explore undeveloped export candidate countries in Korea's food and beverage industry. Node2vec is the method that improves the limit of the structural equivalence representation of the network, which is known to be relatively weak compared to the existing link prediction method based on the number of common neighbors of the network. Therefore, the method is known to show excellent performance in both community detection and structural equivalence of the network. The vector value obtained by embedding the network in this way operates under the condition of a constant length from an arbitrarily designated starting point node. Therefore, it has the advantage that it is easy to apply the sequence of nodes as an input value to the model for downstream tasks such as Logistic Regression, Support Vector Machine, and Random Forest. Based on these features of the Node2vec graph embedding method, this study applied the above method to the international trade information of the Korean food and beverage industry. Through this, we intend to contribute to creating the effect of extensive margin diversification in Korea in the global value chain relationship of the industry. The optimal predictive model derived from the results of this study recorded a precision of 0.95 and a recall of 0.79, and an F1 score of 0.86, showing excellent performance. This performance was shown to be superior to that of the binary classifier based on Logistic Regression set as the baseline model. In the baseline model, a precision of 0.95 and a recall of 0.73 were recorded, and an F1 score of 0.83 was recorded. In addition, the light GBM-based optimal prediction model derived from this study showed superior performance than the link prediction model of previous studies, which is set as a benchmarking model in this study. The predictive model of the previous study recorded only a recall rate of 0.75, but the proposed model of this study showed better performance which recall rate is 0.79. The difference in the performance of the prediction results between benchmarking model and this study model is due to the model learning strategy. In this study, groups were classified by the trade value scale, and prediction models were trained differently for these groups. Specific methods are (1) a method of randomly masking and learning a model for all trades without setting specific conditions for trade value, (2) arbitrarily masking a part of the trades with an average trade value or higher and using the model method, and (3) a method of arbitrarily masking some of the trades with the top 25% or higher trade value and learning the model. As a result of the experiment, it was confirmed that the performance of the model trained by randomly masking some of the trades with the above-average trade value in this method was the best and appeared stably. It was found that most of the results of potential export candidates for Korea derived through the above model appeared appropriate through additional investigation. Combining the above, this study could suggest the practical utility of the link prediction method applying Node2vec and Light GBM. In addition, useful implications could be derived for weight update strategies that can perform better link prediction while training the model. On the other hand, this study also has policy utility because it is applied to trade transactions that have not been performed much in the research related to link prediction based on graph embedding. The results of this study support a rapid response to changes in the global value chain such as the recent US-China trade conflict or Japan's export regulations, and I think that it has sufficient usefulness as a tool for policy decision-making.

• Journal of Intelligence and Information Systems
• /
• v.20 no.4
• /
• pp.1-23
• /
• 2014

Market forecasting aims to estimate the sales volume of a product or service that is sold to consumers for a specific selling period. From the perspective of the enterprise, accurate market forecasting assists in determining the timing of new product introduction, product design, and establishing production plans and marketing strategies that enable a more efficient decision-making process. Moreover, accurate market forecasting enables governments to efficiently establish a national budget organization. This study aims to generate a market growth curve for ICT (information and communication technology) goods using past time series data; categorize products showing similar growth patterns; understand markets in the industry; and forecast the future outlook of such products. This study suggests the useful and meaningful process (or methodology) to identify the market growth pattern with quantitative growth model and data mining algorithm. The study employs the following methodology. At the first stage, past time series data are collected based on the target products or services of categorized industry. The data, such as the volume of sales and domestic consumption for a specific product or service, are collected from the relevant government ministry, the National Statistical Office, and other relevant government organizations. For collected data that may not be analyzed due to the lack of past data and the alteration of code names, data pre-processing work should be performed. At the second stage of this process, an optimal model for market forecasting should be selected. This model can be varied on the basis of the characteristics of each categorized industry. As this study is focused on the ICT industry, which has more frequent new technology appearances resulting in changes of the market structure, Logistic model, Gompertz model, and Bass model are selected. A hybrid model that combines different models can also be considered. The hybrid model considered for use in this study analyzes the size of the market potential through the Logistic and Gompertz models, and then the figures are used for the Bass model. The third stage of this process is to evaluate which model most accurately explains the data. In order to do this, the parameter should be estimated on the basis of the collected past time series data to generate the models' predictive value and calculate the root-mean squared error (RMSE). The model that shows the lowest average RMSE value for every product type is considered as the best model. At the fourth stage of this process, based on the estimated parameter value generated by the best model, a market growth pattern map is constructed with self-organizing map algorithm. A self-organizing map is learning with market pattern parameters for all products or services as input data, and the products or services are organized into an $N{\times}N$ map. The number of clusters increase from 2 to M, depending on the characteristics of the nodes on the map. The clusters are divided into zones, and the clusters with the ability to provide the most meaningful explanation are selected. Based on the final selection of clusters, the boundaries between the nodes are selected and, ultimately, the market growth pattern map is completed. The last step is to determine the final characteristics of the clusters as well as the market growth curve. The average of the market growth pattern parameters in the clusters is taken to be a representative figure. Using this figure, a growth curve is drawn for each cluster, and their characteristics are analyzed. Also, taking into consideration the product types in each cluster, their characteristics can be qualitatively generated. We expect that the process and system that this paper suggests can be used as a tool for forecasting demand in the ICT and other industries.

• Asian Pacific Journal of Cancer Prevention
• /
• v.16 no.11
• /
• pp.4603-4608
• /
• 2015

Background: To determine the potential value of serum tumor markers in predicting pCR (pathological complete response) during neoadjuvant chemotherapy. Materials and Methods: We retrospectively monitored the pro-, mid-, and post-neoadjuvant treatment serum tumor marker concentrations in patients with locally advanced breast cancer (stage II-III) who accepted pre-surgical chemotherapy or chemotherapy in combination with targeted therapy at Fudan University Shanghai Cancer Center between September 2011 and January 2014 and investigated the association of serum tumor marker levels with therapeutic effect. Core needle biopsy samples were assessed using immunohistochemistry (IHC) prior to neoadjuvant treatment to determine hormone receptor, human epidermal growth factor receptor 2(HER2), and proliferation index Ki67 values. In our study, therapeutic response was evaluated by pCR, defined as the disappearance of all invasive cancer cells from excised tissue (including primary lesion and axillary lymph nodes) after completion of chemotherapy. Analysis of variance of repeated measures and receiver operating characteristic (ROC) curves were employed for statistical analysis of the data. Results: A total of 348 patients were recruited in our study after excluding patients with incomplete clinical information. Of these, 106 patients were observed to have acquired pCR status after treatment completion, accounting for approximately 30.5% of study individuals. In addition, 147patients were determined to be Her-2 positive, among whom the pCR rate was 45.6% (69 patients). General linear model analysis (repeated measures analysis of variance) showed that the concentration of cancer antigen (CA) 15-3 increased after neoadjuvant chemotherapy in both pCR and non-pCR groups, and that there were significant differences between the two groups (P=0.008). The areas under the ROC curves (AUCs) of pre-, mid-, and post-treatment CA15-3 concentrations demonstrated low-level predictive value (AUC=0.594, 0.644, 0.621, respectively). No significant differences in carcinoembryonic antigen (CEA) or CA12-5 serum levels were observed between the pCR and non-pCR groups (P=0.196 and 0.693, respectively). No efficient AUC of CEA or CA12-5 concentrations were observed to predict patient response toward neoadjuvant treatment (both less than 0.7), nor were differences between the two groups observed at different time points. We then analyzed the Her-2 positive subset of our cohort. Significant differences in CEA concentrations were identified between the pCR and non-pCR groups (P=0.039), but not in CA15-3 or CA12-5 levels (p=0.092 and 0.89, respectively). None of the ROC curves showed underlying prognostic value, as the AUCs of these three markers were less than 0.7. The ROC-AUCs for the CA12-5 concentrations of inter-and post-neoadjuvant chemotherapy in the estrogen receptor negative HER2 positive subgroup were 0.735 and 0.767, respectively. However, the specificity and sensitivity values were at odds with each other which meant that improving either the sensitivity or specificity would impair the efficiency of the other. Conclusions: Serum tumor markers CA15-3, CA12-5, and CEA might have little clinical significance in predicting neoadjuvant treatment response in locally advanced breast cancer.

• Tuberculosis and Respiratory Diseases
• /
• v.67 no.2
• /
• pp.105-112
• /
• 2009

Background: Solitary pulmonary nodules (SPN) are encountered incidentally in 0.2% of patients who undergo chest X-ray or chest CT. Although SPN has malignant potential, it cannot be treated surgically by biopsy in all patients. The first stage is to determine if patients with SPN require periodic observation and biopsy or resection. An important early step in the management of patients with SPN is to estimate the clinical pretest probability of a malignancy. In every patient with SPN, it is recommended that clinicians estimate the pretest probability of a malignancy either qualitatively using clinical judgment or quantitatively using a validated model. This study examined whether Bayesian analysis or multiple logistic regression analysis is more predictive of the probability of a malignancy in SPN. Methods: From January 2005 to December 2008, this study enrolled 63 participants with SPN at the Kangnam Sacred Hospital. The accuracy of Bayesian analysis and Bayesian analysis with a FDG-PET scan, and Multiple logistic regression analysis was compared retrospectively. The accurate probability of a malignancy in a patient was compared by taking the chest CT and pathology of SPN patients with <30 mm at CXR incidentally. Results: From those participated in study, 27 people (42.9%) were classified as having a malignancy, and 36 people were benign. The result of the malignant estimation by Bayesian analysis was 0.779 (95% confidence interval [CI], 0.657 to 0.874). Using Multiple logistic regression analysis, the result was 0.684 (95% CI, 0.555 to 0.796). This suggests that Bayesian analysis provides a more accurate examination than multiple logistic regression analysis. Conclusion: Bayesian analysis is better than multiple logistic regression analysis in predicting the probability of a malignancy in solitary pulmonary nodules but the difference was not statistically significant.

• Clinical and Experimental Pediatrics
• /
• v.50 no.5
• /
• pp.443-448
• /
• 2007

Purpose : To evaluate the potential prognostic value of the antithrombin-III (AT-III) level in the children with acute lung injury (ALI), we analyzed several early predictive factors of death including AT-III level at the onset of ALI and compared the relative risk of them for mortality. Methods : Over a 18-month period, a total of 198 children were admitted to our pediatric intensive care unit and 21 mechanically ventilated patients met ALI criteria, as defined by American-European consensus conference, i.e., bilateral pulmonary infiltrates and $PaO_2/FiO_2$ lower than 300 without left atrial hypertension. Demographic variables, hemodynamic and respiratory parameters, underlying diseases, as well as Pediatric Risk of Mortality-III (PRISM-III) scores and Lung Injury Score (LIS) at admission were collected. AT-III levels were measured within 3 hours after admission. These variables were compared between survivors and non-survivors and entered into a multiple logistic regression analysis to evaluate their independent prognostic roles. Results : The overall mortality rate was 38.1% (8/21). Non-survivors showed lower age, lower lung compliance, higher PEEP, higher oxygenation index (OI), lower arterial pH, lower $PaO_2/FiO_2$, higher PRISM-III score and LIS, and lower AT-III level. PRISM-III score, LIS, OI and decreased AT-III level (less than 70%) were independently associated with a risk of death and the odds ratio of decreased AT-III level for mortality is 2.75 (95% confidence interval; 1.28-4.12) Conclusion : These results suggest that the decreased level of AT-III is an important prognostic factor in children with ALI and the replacement of AT-III may be considered as an early therapeutic trial.

• Journal of Life Science
• /
• v.24 no.12
• /
• pp.1339-1344
• /
• 2014

Inflammatory bowel disease is an immune disorder associated with chronic mucosal inflammation and severe ulceration in the gastrointestinal tract. Antibodies against proinflammatory cytokines, including TNF${\alpha}$, are currently used as promising therapeutic agents against the disease. Stabilization of the transcript is a crucial post-transcriptional process in the expression of proinflammatory cytokines. In the present study, we assessed the expression and histological distribution of the HuR protein, an important transcript stabilizer, in tissues from experimental animals and patients with Crohn's disease. The total and cytosolic levels of the HuR protein were enhanced in the intestinal epithelia from dextran sodium sulfate (DSS)-treated mice compared to those in control tissues from normal mice. Moreover, the expression of HuR was very high only in the mucosal and glandular epithelium, and the relative localization of the protein was sequestered in the lower parts of the villus during the DSS insult. The expression of HuR was significantly higher in mucosal lesions than in normal-looking areas. Consistent with the data from the animal model, the expression of HuR was confined to the mucosal and glandular epithelium. These results suggest that HuR may contribute to the post-transcriptional regulation of proinflammatory genes during early mucosal insults. More mechanistic investigations are warranted to determine the potential use of HuR as a predictive biomarker or a promising target against IBD.

• Radiation Oncology Journal
• /
• v.25 no.2
• /
• pp.109-117
• /
• 2007

[ $\underline{Purpose}$ ]: To evaluate the incidences and potential predictive factors for symptomatic radiation pneumonitis (SRP) and radiographic pulmonary toxicity (RPT) following adjuvant radiotherapy (RT) for patients with breast cancer. A particular focus was made to correlate RPT with the dose volume histogram (DVH) parameters based on three-dimensional RT planning (3D-RTP) data. $\underline{Materials\;and\;Methods}$: From September 2003 through February 2006, 171 patients with breast cancer were treated with adjuvant RT following breast surgery. A radiation dose of 50.4 Gy was delivered with tangential photon fields on the whole breast or chest wall. A single anterior oblique photon field for supraclavicular (SCL) nodes was added if indicated. Serial follow-up chest radiographs were reviewed by a chest radiologist. Radiation Therapy Oncology Group (RTOG) toxicity criteria were used for grading SRP and a modified World Health Organization (WHO) grading system was used to evaluate RPT. The overall percentage of the ipsilateral lung volume that received ${\geq}15\;Gy\;(V_{15}),\;20\;Gy\;(V_{20})$, and $30\;Gy\;(V_{30})$ and the mean lung dose (MLD) were calculated. We divided the ipsilateral lung into two territories, and defined separate DVH parameters, i.e., $V_{15\;TNGT},\;V_{20\;TNGT},\;V_{30\;TNGT},\;MLD_{TNGT}$, and $V_{15\;SCL},\;V_{20\;SCL},\;V_{30SCL},\;MLD_{SCL}$ to assess the relationship between these parameters and RPT. $\underline{Results}$: Four patients (2.1%) developed SRP (three with grade 3 and one with grade 2, respectively). There was no significant association of SRP with clinical parameters such as, age, pre-existing lung disease, smoking, chemotherapy, hormonal therapy and regional RT. When 137 patients treated with 3D-RTP were evaluated, 13.9% developed RPT in the tangent (TNGT) territory and 49.2% of 59 patients with regional RT developed RPT in the SCL territory. Regional RT (p<0.001) and age (p=0.039) was significantly correlated with RPT. All DVH parameters except for $V_{15\;TNGT}$ showed a significant correlation with RPT (p<0.05). $MLD_{TNGT}$ was a better predictor for RPT for the TNGT territory than $V_{15\;SCL}$ for the SCL territory. $\underline{Conclusion}$: The incidence of SRP was acceptable with the RT technique that was used. Age and regional RT were significant factors to predict RPT. The DVH parameter was good predictor for RPT for the SCL territory while $MLD_{TNGT}$ was a better predictor for RPT for the TNGT territory.

• Journal of Life Science
• /
• v.21 no.11
• /
• pp.1518-1525
• /
• 2011

Sorafenib is the only approved systemic, therapeutic agent for hepatocellular carcinoma (HCC). The use of Ginseng Extract (GE) in cancer patients is growing worldwide; however, drug interaction between sorafenib and GE has not been illuminated. Four different human cancer cell lines including HepG2 were used and immunocompetent mice were implanted subcutaneously with a mouse HCC cell line. Treatment with low dose GE stimulated cell growth, while a high dose inhibited growth. pERK (phosphorylation of extracellular signal-regulated kinase) was concomitantly increased and decreased respective of different doses of GE. Antitumoral effect of sorafenib decreased in non-proliferating phase cells but was sensitized after low dose GE (LDG) treatment. PD98059 (ERK phosphorylation inhibitor) efficiently blocked ERK phosphorylation, resulting in loss of sorafenib sensitization even after LDG treatment. In the HCC mouse model, LDG alone slightly increased tumor size while sorafenib alone significantly decreased it. However, a combination of LDG and sorafenib significantly decreased tumor size compared with sorafenib alone. Increase of pERK was observed in some normal mice organs and mild inflammatory change was observed in some of these organs, suggesting pERK activation by LDG may cause unexpected toxicity in normal cells. GE, dose-dependently, induced stimulation or inhibition in some human cancer cell lines. Combinational use of GE and sorafenib possibly potentiated an antitumoral response to sorafenib. pERK level has been provided as a potential predictive marker for sorafenib. Our result may suggest GE's dual effects in relation to pERK level in HCC cancer cell lines, and that certain doses of GE can sensitize sorafenib.

• Radiation Oncology Journal
• /
• v.15 no.1
• /
• pp.57-64
• /
• 1997

Purpose : Before we report the results of curative radiotherapy in cervix cancer patients, we review the significance and safety of our dose specification methods in the brachytherapy system to have the insight of the potential Predictive value of doses at specified points. Matersials and Methods : We analyze the 리5 cases of cervix cancer patients treated with intracavitary brachytherapy in the lateral simulation film we draw the isodose curve and observe the absorbed dose rate of point A, the reference point of bladder(SBD) and rectum(SRD). In the sagittal view of Pelvic MRI film we demarcate the tumor volume(TV) and determine whether the prescription dose curve of point A covers the tumor volume adequately by drawing the isodose curve as correctly as possible. Also we estimate the maximum Point dose of bladder(MBD) and rectum(MRD) and calculate the inclusion area where the absorbed dose rate is higher than that of point A in the bladder(HBV) and rectum(HRV), respectively. Results : Of forty-five cases, the isodose curve of point A seems to cover tumor volume optimally in only 24(53%). The optimal tumor coverage seems to be associated not with the stage of the disease but with the tumor volume. There is no statistically significant association between SBD/SRD and MBD/MRD, respectively. SRD has statistically marginally significant association with HRV, while TV has statistically significant association with HBV and HRV. Conclusion : Our current treatment calculation methods seem to have the defect in the aspects of the nonoptimal coverage of the bulky tumor and the inappropriate estimation of bladder dose. We therefore need to modify the applicator geometry to optimize the dose distribution at the position of lower tandem source. Also it appears that the position of the bladder in relation to the applicators needs to be defined individually to define 'hot spots'.

• Journal of Korea Water Resources Association
• /
• v.54 no.spc1
• /
• pp.1143-1154
• /
• 2021

The selection of activation function has a great influence on the groundwater level prediction performance of artificial neural network (ANN) model. In this study, five activation functions were applied to ANN model for two groundwater level observation wells in the middle mountainous area of the Pyoseon watershed in Jeju Island. The results of the prediction of the groundwater level were compared and analyzed, and the optimal activation function was derived. In addition, the results of LSTM model, which is a widely used recurrent neural network model, were compared and analyzed with the results of the ANN models with each activation function. As a result, ELU and Leaky ReLU functions were derived as the optimal activation functions for the prediction of the groundwater level for observation well with relatively large fluctuations in groundwater level and for observation well with relatively small fluctuations, respectively. On the other hand, sigmoid function had the lowest predictive performance among the five activation functions for training period, and produced inappropriate results in peak and lowest groundwater level prediction. The ANN-ELU and ANN-Leaky ReLU models showed groundwater level prediction performance comparable to that of the LSTM model, and thus had sufficient potential for application. The methods and results of this study can be usefully used in other studies.