• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.1-25
• /
• 2020

In this paper, we suggest an application system architecture which provides accurate, fast and efficient automatic gasometer reading function. The system captures gasometer image using mobile device camera, transmits the image to a cloud server on top of private LTE network, and analyzes the image to extract character information of device ID and gas usage amount by selective optical character recognition based on deep learning technology. In general, there are many types of character in an image and optical character recognition technology extracts all character information in an image. But some applications need to ignore non-of-interest types of character and only have to focus on some specific types of characters. For an example of the application, automatic gasometer reading system only need to extract device ID and gas usage amount character information from gasometer images to send bill to users. Non-of-interest character strings, such as device type, manufacturer, manufacturing date, specification and etc., are not valuable information to the application. Thus, the application have to analyze point of interest region and specific types of characters to extract valuable information only. We adopted CNN (Convolutional Neural Network) based object detection and CRNN (Convolutional Recurrent Neural Network) technology for selective optical character recognition which only analyze point of interest region for selective character information extraction. We build up 3 neural networks for the application system. The first is a convolutional neural network which detects point of interest region of gas usage amount and device ID information character strings, the second is another convolutional neural network which transforms spatial information of point of interest region to spatial sequential feature vectors, and the third is bi-directional long short term memory network which converts spatial sequential information to character strings using time-series analysis mapping from feature vectors to character strings. In this research, point of interest character strings are device ID and gas usage amount. Device ID consists of 12 arabic character strings and gas usage amount consists of 4 ~ 5 arabic character strings. All system components are implemented in Amazon Web Service Cloud with Intel Zeon E5-2686 v4 CPU and NVidia TESLA V100 GPU. The system architecture adopts master-lave processing structure for efficient and fast parallel processing coping with about 700,000 requests per day. Mobile device captures gasometer image and transmits to master process in AWS cloud. Master process runs on Intel Zeon CPU and pushes reading request from mobile device to an input queue with FIFO (First In First Out) structure. Slave process consists of 3 types of deep neural networks which conduct character recognition process and runs on NVidia GPU module. Slave process is always polling the input queue to get recognition request. If there are some requests from master process in the input queue, slave process converts the image in the input queue to device ID character string, gas usage amount character string and position information of the strings, returns the information to output queue, and switch to idle mode to poll the input queue. Master process gets final information form the output queue and delivers the information to the mobile device. We used total 27,120 gasometer images for training, validation and testing of 3 types of deep neural network. 22,985 images were used for training and validation, 4,135 images were used for testing. We randomly splitted 22,985 images with 8:2 ratio for training and validation respectively for each training epoch. 4,135 test image were categorized into 5 types (Normal, noise, reflex, scale and slant). Normal data is clean image data, noise means image with noise signal, relfex means image with light reflection in gasometer region, scale means images with small object size due to long-distance capturing and slant means images which is not horizontally flat. Final character string recognition accuracies for device ID and gas usage amount of normal data are 0.960 and 0.864 respectively.

• Journal of Preventive Medicine and Public Health
• /
• v.30 no.3 s.58
• /
• pp.630-642
• /
• 1997

The purpose of this study was to examine the mortality risk associated with cognitive impairment among the rural elderly. The subjective of study was 558 of 'A Study on the Depression and Cognitive Impairment in the Rural Elderly' of Jung Ae Rhee and Hyang Gyun Jung's study(1993). Cognitive impairment and other social and health factors were assessed in 558 elderly rural community residents. For this study, a Korean version of the Mini-Mental State Examination(MMSEK) was used as a global indicator of cognitive functioning. And mortality risk factors for each cognitive impairment subgroup were identified by univariate and multivariate Cox regression analysis. At baseline 22.6% of the sample were mildly impaired and 14.2% were severely impaired. As the age increased, the cognitive function was more impaired. Sexual difference was existed in the cognitive function level. Also the variables such as smoking habits, physical disorders had the significant relationship with cognitive function impairment. Across a 3-year observation period the mortality rate was 8.5% for the cognitively unimpaired, 11.1% for the mildly impaired, and 16.5% for the severly impaired respendents. And the survival probability was .92 for the cognitively unimpaired, .90 for the mildly impaired, and .86 for the severly impaired respondents. Compared to survival curve for the cognitively unimpaired group, each survival curve for the mildly and the severely impaired group was not significantly different. When adjustments models were not made for the effects of other health and social covariates, each hazard ratio of death of mildly and severely impaired persons was not significantly different as compared with the cognitively unimpaired. But, as MMSEK score increased, significantly hazard ratio of death decreased. Employing Cox univariate proportional hazards model, statistically other significant variables were age, monthly income, smoking habits, physical disorders. Also when adjustments were made for the effects of other health and social covariates, there was no difference in hazard ratio of death between those with severe or mild impairment and unimpaired persons. And as MMSEK score increased, significantly hazard ratio of death did not decrease. Employing Cox multivariate proportional hazards model, statistically other significant variables were age, monthly income, physical disorders. Employing Cox multivariate proportional hazards model by sex, at men and women statistically significant variable was only age. For both men and women, also cognitive impairment was not a significant risk factor. Other investigators have found that cognitive impairment is a significant predictor of mortality. But we didn't find that it is a significant predictor of mortality. Even though the conclusions of our study were not related to cognitive impairment and mortality, early detection of impaired cognition and attention to associated health problems could improve the quality of life of these older adults and perhaps extend their survival.

• Childhood Kidney Diseases
• /
• v.7 no.2
• /
• pp.142-149
• /
• 2003

Purpose : The urinary mass screening program for the detection of urinary abnormalities in school aged population has been performed in Seoul since 1981. Nation-wide urinary mass screening program was also performed since 1998. The aim of this study was to analyze the cause and nature of isolated proteinuria detected by chance on the urinary mass screening test in Busan and Kyungsangnam-do Province Methods : The medical records of 44 cases of isolated proteinuria detected by chance on the urinary mass screening test in Busan and Kyungsangnam-do Province, and evaluated for urinary abnormalities at the pediatrics outpatients renal clinics of Busan Paik Hospital from April 2002 to August 2003 were reviewed prospectively. Results : The cause and incidence of isolated proteinuria were as follows; transient proteinuria 4 cases(9.1%), orthostatic proteinuria 36 cases(81.8%) and persistent proteinuria 4 cases (9.1%). The total protein amount of the 24 hour urine were $121.0{\pm}136.4\;mg$ in transient proteinuria, $179.1{\pm}130.0\;mg$ in orthostatic proteinuria and $1532.8{\pm}982.5\;mg$ in persistent proteinuria. In the orthostatic proteinuria group, the total protein amount of the 24 hour urine was in the range of 40-616 mg. Spot urine protein/creatinine ratio(PCR) were $0.10{\pm}0.01$ in transient proteinuria, $0.61{\pm}0.61$ in orthostatic proteinuria and $4.35{\pm}4.04$ in persistent proteinuria. In the orthostatic proteinuria group, spot me PCR was in the range of 0.09-2.32. Renal biopsy was peformed in 4 children of the persisitent proteinuria group. They showed minimal change in 1 case, membranoproliferatiye glomerulonephritis in 2 cases and secondary renal amyloidosis in 1 case. Conclusion : The majority of isolated proteinuria which was detected by chance on school urinary mass screening were transient or orthostatic proteinuria. Even though the incidence of persistent proteinuria was much lower, it is necessary to take care of these children regularly and continuously, because persistent proteinuria itself is a useful marker of the progressive renal problems.

• The Korean Journal of Nuclear Medicine
• /
• v.39 no.1
• /
• pp.34-43
• /
• 2005

Purpose: $^{99m}Tc$-sestamibi(MIBI) and $^{99m}Tc$-tetrofosmin have been used as substrates for P-glycoprotein (Pgp) and multidrug resistance associated protein (MRP), which are closely associated with multidrug resistance of the tumors. To understand different handling of radiotracers in cancer cell lines expressing Pgp and MRP, we compared cellular uptakes of $^{99m}Tc$-MIBI and $^{99m}Tc$-tetrofosmin. The effects of cyclosporin A (CsA), well-known multidrug resistant reversing agent, on the uptake of both tracers were also compared. Materials and Methods: HCT15/CL02 human colorectal cancer cells for Pgp expressing cells, and human non-small cell lung cancer A549 cells for MRP expressing cells, were used for in vitro and in vivo studies. RT-PCR, western blot analysis and immunohistochemistry were used for detection of Pgp and MRP. MDR-reversal effect with CsA was evaluated at different drug concentrations after incubation with MIBI or tetrofosmin. Radioactivities of supernatant and pellet were measured with gamma well counter. Tumoral uptake of the tracers were measured from tumor bearing nude mice treated with or without CsA. Results: RT-PCR, western blot analysis of the cells and irnrnunochemical staining revealed selective expression of Pgp and MRP for HCY15/CL02 and A549 cells, respectively. There were no significant difference in cellular uptakes of both tracers in HCT15/CL02 cells, but MIBI uptake was slightly higher than that of tetrofosmin in A549 cells. Co-incubation with CsA resulted in a increase in cellular uptakes of MIBI and tetrofosmin. Uptake of MIBI or tetrofosmin in HCT15/CL02 cells was increased by 10- and 2.4-fold, and by 7.5 and 6.3-fold in A549 cells, respectively. Percentage increase of MIBI was higher than that of tetrofosmin with CsA for both cells (p<0.05). In vivo biodistribution study showed that MIBI (114% at 10 min, 257% at 60 min, 396% at 240 min) and tetrofosmin uptake (110% at 10 min, 205% at 60 min, 410% at 240 min) were progressively increased by the time, up to 240 min with CsA. But increases in tumoral uptake were not significantly different between MIBI and tetrofosmin for both tumors. Conclusion: MIBI seems to be a better tracer than tetrofosmin for evaluating MDR reversal effect of the modulators in vitro, but these differences were not evident in vivo tumoral uptake. Both MIBI and tetrofosmin seem to be suitable tracers for imaging Pgp- and MRP-mediated drug resistance in tumors.

• Journal of Food Hygiene and Safety
• /
• v.21 no.4
• /
• pp.231-237
• /
• 2006

Nowadays the teabag is worldwide used for various products including green tea, tea, coffee, etc. since it is convenient for use. In case of outer packaging printed, however, there is a possibility that the plasticizers which is used for improvement in adhesiveness of printing ink may shift to inner tea bag. In this study, in order to monitor residual levels of plasticizers in teabags, we have established the simultaneous analysis method of 9 phthalates and 7 adipates plasticizers using gas chromatography (GC). These compounds were also confirmed using gas chromatography-mass spectrometry (GC-MSD). The recoveries of plasticizers analyzed by GC ranged from 82.7% to 104.6% with coefficient of variation of $0.6\sim2.7%$ and the correlation coefficients of each plasticizer was $0.9991\sim0.9999$. Therefore this simultaneous analysis method was showed excellent reproducibility and linearity. And limit of detection (LOD) and limit of quantitation (LOQ) on individual plasticizer were $0.1\sim3.5\;ppm\;and\;0.3\sim11.5\;ppm$ respectively. When 143 commercial products of teabag were monitored, no plasticizers analysed were detected in filter of teabag products. The migration into $95^{\circ}C$ water as food was also examined and the 16 plasticizers are not detected. In addition we carried out analysis of heavy metals, lead (Pb), cadmium (Cd), arsenic (As) and aluminum (Al) in teabag filters using ICP/AES. $Trace\sim23{\mu}g$ Pb per teabag and $0.6\sim1718{\mu}g$ Al per teabag were detected in materials of samples and Cd and As are detected less than LOQ (0.05 ppm). The migration levels of Pb and Al from teabag filter to $95^{\circ}C$ water were upto $11.5{\mu}g\;and\;20.8{\mu}g$ per teabag, respectively and Cd and As were not detected in exudate water of all samples. Collectively, these results suggest that there is no safety concern from using teabag filter.

• Journal of Food Hygiene and Safety
• /
• v.32 no.6
• /
• pp.470-476
• /
• 2017

The purpose of this study was to investigate and evaluate the safety of the grains, nut products, beans and oilseeds being sold in Gyeonggi province by analyzing mycotoxins. A multi-mycotoxins analysis method based on LC-MS/MS was validated and applied for the determination of eight mycotoxins, including aflatoxins ($B_1$, $B_2$, $G_1$ and $G_2$), fumonisins ($B_1$, $B_2$), zearalenone and ochratoxcin A in 134 samples. The limit of detection (LOD) and limit of quantitation (LOQ) for the eight mycotoxins ranged from 0.14 to $8.25{\mu}g/kg$ and from 1.08 to $7.21{\mu}g/kg$, respectively. Recovery rates of mycotoxins were determined in the range of 61.1 to 97.5% with RSD of 1.0~14.5% (n=3). Fumonisin $B_1$, $B_2$, zearalenone, and ochratoxin A were detected in 22 samples, indicating that 27% of grains, 12.5% of beans and 11.8% of oilseeds were contaminated. Fumonisin and zearalenone were detected simultaneously in 2 adlays and 3 sorghums. Fumonisin $B_1$ and $B_2$ were detected simultaneously in most samples whereas fumonisin $B_1$ was detected in 1 adlay, 1 millet and 1 sesame sample. The average detected amount of fumonisin was $49.3{\mu}g/kg$ and $10.1{\mu}g/kg$ for grains and oilseeds, respectively. The average detected amount of zearalenone was $1.9{\mu}g/kg$ and $1.5{\mu}g/kg$ for grains and beans, respectively. In addition, the average amount of ochratoxin A was $0.08{\mu}g/kg$ for grains. The calculated exposure amounts of fumonisin, zeralenone and ochratoxin A for grains, beans and oilseeds were below the PMTDI/PTWI.

• Korean Journal of Psychosomatic Medicine
• /
• v.27 no.1
• /
• pp.50-59
• /
• 2019

Objectives : Many of the patients with type 2 diabetes are associated with sleep problems, and the rate of insomnia is known to be higher in the general population. The aims of this study were to know the frequency and clnical characteristics of insomnia, and related variables to insomnia in patients diagnosed with type 2 diabetes. Methods : For 99 patients from 18 to 80 years of age (65 males and 34 females) with type 2 diabetes, interviews were performed. Total sleep time and sleep latency was evaluated. Insomnia was evaluated using the Korean Version of the Insomnia Severity Index (ISI-K). Severity of depressive symptoms were evaluted using the Korean version of the Hamilton Depression Scale (K-HDRM). According to the cutoff score of 15.5 on the ISI-K, subjects were divided into the group of type 2 diabetics with insomnia (N=34) and those without insomnia (N=65) at first, and then statistically analyzed. Results : TInsomnia could be found in 34.34% of type 2 diabetics. Type 2 diabetics with insomnia had significantly more single or divorced (respectively 11.8%, p<0.05), higher total scores of the K-HDRS ($11.76{\pm}5.52$, p<0.001), shorter total sleep time ($5.35{\pm}2.00hours$, p<0.001), and longer sleep latency ($50.29{\pm}33.80minutes$, p<0.001). The all item scores of the ISI-K in type 2 diabetics with insomnia were significantly higher than those in type 2 diabetics without insomnia, that is, total ($18.38{\pm}2.69$), A1 (Initial insomnia) ($2.97{\pm}0.76$), A2 (Middle insomnia) ($3.06{\pm}0.69$), A3 (Terminal insomnia) ($2.76{\pm}0.61$), B (Satisfaction) ($3.18{\pm}0.72$), C (Interference) ($2.09{\pm}0.97$), D (Noticeability) ($2.12{\pm}1.09$) and E (Distress) ($2.21{\pm}0.81$) (respectively p<0.001). Variables associated with insomnia in type 2 diabetics were as following. Age had significant negative correlation with A3 items of the ISI-K (${\beta}=-0.241$, p<0.05). Total scores of the K-HDRS had significant positive correlation, while total sleep time had significant negative correlation with all items of the ISI-K (respectively p<0.05). Sleep latency had significant positive correlation with total,, A1, B and E item scores of the ISI-K (respectively p<0.05). Conclusions : Insomnia was found in about 1/3 of type 2 diabetics. According to the presence of insomnia, clinical characteristics including sleep quality as well as quantity seemed to be different. Because depression seemed to be correlated with insomnia, clinicians should pay attention to early detection and intervention of depression among type 2 diabetics.

• Korean Journal of Environmental Agriculture
• /
• v.37 no.4
• /
• pp.291-301
• /
• 2018

BACKGROUND: Rice is one of the main sources for inorganic arsenic among the consumed crops in the world population's diet. Arsenic is classified into Group 1 as it is carcinogenic for humans, according to the IARC. This study was carried out to assess dietary exposure risk of inorganic arsenic in husked rice and polished rice to the Korean population health. METHODS AND RESULTS: Total arsenic was determined using microwave device and ICP-MS. Inorganic arsenic was determined by ICP-MS coupled with HPLC system. The HPLC-ICP-MS analysis was optimized based on the limit of detection, limit of quantitation, and recovery ratio to be $0.73-1.24{\mu}g/kg$, $2.41-4.09{\mu}g/kg$, and 96.5-98.9%, respectively. The inorganic arsenic concentrations of daily exposure (included in body weight) were $4.97{\times}10^{-3}$ (${\geq}20$ years old) $-1.36{\times}10^{-2}$ (${\leq}2$ years old) ${\mu}g/kg\;b.w./day$ (PTWI 0.23-0.63%) by the husked rice, and $1.39{\times}10^{-1}$ (${\geq}20$ years old) $-3.21{\times}10^{-1}$ (${\leq}2$ years old) ${\mu}g/kg\;b.w./day$ (PTWI 6.47-15.00%) by the polished rice. CONCLUSION: The levels of overall exposure to total and inorganic arsenic by the husked and polished rice were far lower than the recommended levels of The Joint FAO/WHO Expert Committee on Food Additives (JECFA), indicating of little possibility of risk.

• Journal of Food Hygiene and Safety
• /
• v.34 no.2
• /
• pp.124-134
• /
• 2019

An analytical method was developed for the determination of broflanilide and its metabolites in agricultural products. Sample preparation was conducted using the QuEChERS (Quick, Easy, Cheap, Effective, Rugged and Safe) method and LC-MS/MS (liquid chromatograph-tandem mass spectrometer). The analytes were extracted with acetonitrile and cleaned up using d-SPE (dispersive solid phase extraction) sorbents such as anhydrous magnesium sulfate, primary secondary amine (PSA) and octadecyl ($C_{18}$). The limit of detection (LOD) and quantification (LOQ) were 0.004 and 0.01 mg/kg, respectively. The recovery results for broflanilide, DM-8007 and S(PFP-OH)-8007 ranged between 90.7 to 113.7%, 88.2 to 109.7% and 79.8 to 97.8% at different concentration levels (LOQ, 10LOQ, 50LOQ) with relative standard deviation (RSD) less than 8.8%. The inter-laboratory study recovery results for broflanilide and DM-8007 and S (PFP-OH)-8007 ranged between 86.3 to 109.1%, 87.8 to 109.7% and 78.8 to 102.1%, and RSD values were also below 21%. All values were consistent with the criteria ranges requested in the Codex guidelines (CAC/GL 40-1993, 2003) and the Food and Drug Safety Evaluation guidelines (2016). Therefore, the proposed analytical method was accurate, effective and sensitive for broflanilide determination in agricultural commodities.

• Journal of Intelligence and Information Systems
• /
• v.26 no.4
• /
• pp.1-25
• /
• 2020

Sentiment Analysis (SA) is a Natural Language Processing (NLP) task that analyzes the sentiments consumers or the public feel about an arbitrary object from written texts. Furthermore, Aspect-Based Sentiment Analysis (ABSA) is a fine-grained analysis of the sentiments towards each aspect of an object. Since having a more practical value in terms of business, ABSA is drawing attention from both academic and industrial organizations. When there is a review that says "The restaurant is expensive but the food is really fantastic", for example, the general SA evaluates the overall sentiment towards the 'restaurant' as 'positive', while ABSA identifies the restaurant's aspect 'price' as 'negative' and 'food' aspect as 'positive'. Thus, ABSA enables a more specific and effective marketing strategy. In order to perform ABSA, it is necessary to identify what are the aspect terms or aspect categories included in the text, and judge the sentiments towards them. Accordingly, there exist four main areas in ABSA; aspect term extraction, aspect category detection, Aspect Term Sentiment Classification (ATSC), and Aspect Category Sentiment Classification (ACSC). It is usually conducted by extracting aspect terms and then performing ATSC to analyze sentiments for the given aspect terms, or by extracting aspect categories and then performing ACSC to analyze sentiments for the given aspect category. Here, an aspect category is expressed in one or more aspect terms, or indirectly inferred by other words. In the preceding example sentence, 'price' and 'food' are both aspect categories, and the aspect category 'food' is expressed by the aspect term 'food' included in the review. If the review sentence includes 'pasta', 'steak', or 'grilled chicken special', these can all be aspect terms for the aspect category 'food'. As such, an aspect category referred to by one or more specific aspect terms is called an explicit aspect. On the other hand, the aspect category like 'price', which does not have any specific aspect terms but can be indirectly guessed with an emotional word 'expensive,' is called an implicit aspect. So far, the 'aspect category' has been used to avoid confusion about 'aspect term'. From now on, we will consider 'aspect category' and 'aspect' as the same concept and use the word 'aspect' more for convenience. And one thing to note is that ATSC analyzes the sentiment towards given aspect terms, so it deals only with explicit aspects, and ACSC treats not only explicit aspects but also implicit aspects. This study seeks to find answers to the following issues ignored in the previous studies when applying the BERT pre-trained language model to ACSC and derives superior ACSC models. First, is it more effective to reflect the output vector of tokens for aspect categories than to use only the final output vector of [CLS] token as a classification vector? Second, is there any performance difference between QA (Question Answering) and NLI (Natural Language Inference) types in the sentence-pair configuration of input data? Third, is there any performance difference according to the order of sentence including aspect category in the QA or NLI type sentence-pair configuration of input data? To achieve these research objectives, we implemented 12 ACSC models and conducted experiments on 4 English benchmark datasets. As a result, ACSC models that provide performance beyond the existing studies without expanding the training dataset were derived. In addition, it was found that it is more effective to reflect the output vector of the aspect category token than to use only the output vector for the [CLS] token as a classification vector. It was also found that QA type input generally provides better performance than NLI, and the order of the sentence with the aspect category in QA type is irrelevant with performance. There may be some differences depending on the characteristics of the dataset, but when using NLI type sentence-pair input, placing the sentence containing the aspect category second seems to provide better performance. The new methodology for designing the ACSC model used in this study could be similarly applied to other studies such as ATSC.