1.Occupation classification model based on DistilKoBERT: using the 5th and 6th Korean Working Condition Surveys
Tae-Yeon KIM ; Seong-Uk BAEK ; Myeong-Hun LIM ; Byungyoon YUN ; Domyung PAEK ; Kyung Ehi ZOH ; Kanwoo YOUN ; Yun Keun LEE ; Yangho KIM ; Jungwon KIM ; Eunsuk CHOI ; Mo-Yeol KANG ; YoonHo CHO ; Kyung-Eun LEE ; Juho SIM ; Juyeon OH ; Heejoo PARK ; Jian LEE ; Jong-Uk WON ; Yu-Min LEE ; Jin-Ha YOON
Annals of Occupational and Environmental Medicine 2024;36(1):e19-
Accurate occupation classification is essential in various fields, including policy development and epidemiological studies. This study aims to develop an occupation classification model based on DistilKoBERT. This study used data from the 5th and 6th Korean Working Conditions Surveys conducted in 2017 and 2020, respectively. A total of 99,665 survey participants, who were nationally representative of Korean workers, were included. We used natural language responses regarding their job responsibilities and occupational codes based on the Korean Standard Classification of Occupations (7th version, 3-digit codes). The dataset was randomly split into training and test datasets in a ratio of 7:3. The occupation classification model based on DistilKoBERT was fine-tuned using the training dataset, and the model was evaluated using the test dataset. The accuracy, precision, recall, and F1 score were calculated as evaluation metrics. The final model, which classified 28,996 survey participants in the test dataset into 142 occupational codes, exhibited an accuracy of 84.44%. For the evaluation metrics, the precision, recall, and F1 score of the model, calculated by weighting based on the sample size, were 0.83, 0.84, and 0.83, respectively. The model demonstrated high precision in the classification of service and sales workers yet exhibited low precision in the classification of managers. In addition, it displayed high precision in classifying occupations prominently represented in the training dataset. This study developed an occupation classification system based on DistilKoBERT, which demonstrated reasonable performance. Despite further efforts to enhance the classification accuracy, this automated occupation classification model holds promise for advancing epidemiological studies in the fields of occupational safety and health.
2.Occupation classification model based on DistilKoBERT: using the 5th and 6th Korean Working Condition Surveys
Tae-Yeon KIM ; Seong-Uk BAEK ; Myeong-Hun LIM ; Byungyoon YUN ; Domyung PAEK ; Kyung Ehi ZOH ; Kanwoo YOUN ; Yun Keun LEE ; Yangho KIM ; Jungwon KIM ; Eunsuk CHOI ; Mo-Yeol KANG ; YoonHo CHO ; Kyung-Eun LEE ; Juho SIM ; Juyeon OH ; Heejoo PARK ; Jian LEE ; Jong-Uk WON ; Yu-Min LEE ; Jin-Ha YOON
Annals of Occupational and Environmental Medicine 2024;36(1):e19-
Accurate occupation classification is essential in various fields, including policy development and epidemiological studies. This study aims to develop an occupation classification model based on DistilKoBERT. This study used data from the 5th and 6th Korean Working Conditions Surveys conducted in 2017 and 2020, respectively. A total of 99,665 survey participants, who were nationally representative of Korean workers, were included. We used natural language responses regarding their job responsibilities and occupational codes based on the Korean Standard Classification of Occupations (7th version, 3-digit codes). The dataset was randomly split into training and test datasets in a ratio of 7:3. The occupation classification model based on DistilKoBERT was fine-tuned using the training dataset, and the model was evaluated using the test dataset. The accuracy, precision, recall, and F1 score were calculated as evaluation metrics. The final model, which classified 28,996 survey participants in the test dataset into 142 occupational codes, exhibited an accuracy of 84.44%. For the evaluation metrics, the precision, recall, and F1 score of the model, calculated by weighting based on the sample size, were 0.83, 0.84, and 0.83, respectively. The model demonstrated high precision in the classification of service and sales workers yet exhibited low precision in the classification of managers. In addition, it displayed high precision in classifying occupations prominently represented in the training dataset. This study developed an occupation classification system based on DistilKoBERT, which demonstrated reasonable performance. Despite further efforts to enhance the classification accuracy, this automated occupation classification model holds promise for advancing epidemiological studies in the fields of occupational safety and health.
3.Occupation classification model based on DistilKoBERT: using the 5th and 6th Korean Working Condition Surveys
Tae-Yeon KIM ; Seong-Uk BAEK ; Myeong-Hun LIM ; Byungyoon YUN ; Domyung PAEK ; Kyung Ehi ZOH ; Kanwoo YOUN ; Yun Keun LEE ; Yangho KIM ; Jungwon KIM ; Eunsuk CHOI ; Mo-Yeol KANG ; YoonHo CHO ; Kyung-Eun LEE ; Juho SIM ; Juyeon OH ; Heejoo PARK ; Jian LEE ; Jong-Uk WON ; Yu-Min LEE ; Jin-Ha YOON
Annals of Occupational and Environmental Medicine 2024;36(1):e19-
Accurate occupation classification is essential in various fields, including policy development and epidemiological studies. This study aims to develop an occupation classification model based on DistilKoBERT. This study used data from the 5th and 6th Korean Working Conditions Surveys conducted in 2017 and 2020, respectively. A total of 99,665 survey participants, who were nationally representative of Korean workers, were included. We used natural language responses regarding their job responsibilities and occupational codes based on the Korean Standard Classification of Occupations (7th version, 3-digit codes). The dataset was randomly split into training and test datasets in a ratio of 7:3. The occupation classification model based on DistilKoBERT was fine-tuned using the training dataset, and the model was evaluated using the test dataset. The accuracy, precision, recall, and F1 score were calculated as evaluation metrics. The final model, which classified 28,996 survey participants in the test dataset into 142 occupational codes, exhibited an accuracy of 84.44%. For the evaluation metrics, the precision, recall, and F1 score of the model, calculated by weighting based on the sample size, were 0.83, 0.84, and 0.83, respectively. The model demonstrated high precision in the classification of service and sales workers yet exhibited low precision in the classification of managers. In addition, it displayed high precision in classifying occupations prominently represented in the training dataset. This study developed an occupation classification system based on DistilKoBERT, which demonstrated reasonable performance. Despite further efforts to enhance the classification accuracy, this automated occupation classification model holds promise for advancing epidemiological studies in the fields of occupational safety and health.
4.Analysis of Retinal Nerve Fiber Layer Thickness in Pediatric Migraine Using Optical Coherence Tomography
Ha Eun SIM ; Su Jeong YOU ; Jin CHOI
Journal of the Korean Ophthalmological Society 2023;64(10):951-956
Purpose:
We investigated disparities in retinal nerve fiber layer (RNFL) thickness surrounding the optic disc among pediatric migraine patients based on the presence of a concomitant visual aura.
Methods:
We used optical coherence tomography to measure RNFL thickness around the optic disc in children with migraines and no organic diseases, along with a control group of healthy children. Differences in RNFL thickness were analyzed based on the pattern of prognostic symptoms. Additionally, we investigated the effects of migraine duration and frequency on RNFL thickness.
Results:
In total, 45 children with migraine and 30 healthy controls were enrolled in this study. Among the children with migraine, 15 experienced visual aura, whereas 30 did not exhibit visual aura. Visual prognostic symptoms included blurred vision, scotoma, and scintillating scotoma. Seven children presented with blurred vision and scotoma, whereas eight reported scintillating scotomas. The mean RNFL thickness around the optic disc was significantly lower in patients with blurred vision and scotomas (89.19 ± 7.99 μm) compared with the control group (98.79 ± 8.49 μm), patients without visual aura (100.55 ± 6.43 μm), and patients with scintillating scotoma (102.21 ± 10.13 μm, p < 0.05).
Conclusions
RNFL thickness around the optic disc was significantly reduced in children with migraine who reported symptoms of blurred vision and scotoma.
5.Verification of the Reliability and Validity of the Korean Pediatric Patient Classification System and Estimation of Nursing Time Conversion Index
Mi Young SIM ; Ji Sun PARK ; Mi Kyung KWON ; Suk Hee SONG ; Ye Seul KIM ; Min Seo KANG ; Shin Ae LEE ; Eun Seok CHOI ; Eun Ju HA
Journal of Korean Clinical Nursing Research 2022;28(2):185-197
Purpose:
This study was performed to verify reliability and validity of the Korean Pediatric Patient Classification System (KPPCS) and estimate nursing time conversion index.
Methods:
The study was conducted in 9 children’s hospital which included various areas and size of bed settings. To verify intraclass correlation reliability and construct validity, staff nurses and nurse managers of 21 wards classified 575 patients according to KPPCS comparing differences by age, days of stay, type of stay and medical department. Direct and indirect nursing time of 575 patients were measured by 284 nursing staffs by stopwatch observation and self reports for 24 hours.
Results:
KPPCS has 12 categories, 55 nursing activities and 80 criterions. High agreement among nurses (r=.91, p<.001) suggested substantial reliability. Construct validity was verified by comparing differences in age, days of stay, type of stay and medical department (p<.05). The correlation of nursing time and classification score was also statistically significant (r=.59, p<.001). The nursing time conversion index was 10.78 minutes per 1 classification score. The entire patient group were classified to four groups using KPPCS.
Conclusion
The findings suggest that KPPCS would be a useful tool for estimating nursing demands related to the complexity of pediatric patients.
6.Scientific Publications on Thyroid Ultrasound between 2001 and 2020: Differences in Research Characteristics by Disciplines
Won Chul SHIN ; Chae Woon LEE ; Jiyeon HA ; Kyoung Ja LIM ; Young Lan SEO ; Eun Joo YUN ; Dae Young YOON
Korean Journal of Radiology 2022;23(8):835-845
Objective:
To analyze the characteristics and trends of scientific publications on thyroid ultrasound (US) from 2001 to 2020, specifically examining the differences among disciplines.
Materials and Methods:
The MEDLINE database was searched for scientific articles on thyroid US published between 2001 and 2020 using the PubMed online service. The evaluated parameters included year of publication, type of document, topic, funding, first author’s specialty, journal name, subject category, impact factor, and quartile ranking of the publishing journal, country, and language. Relationships between the first author’s specialty (radiology, internal medicine, surgery, otorhinolaryngology, and miscellaneous) and other parameters were analyzed.
Results:
A total of 2917 thyroid US publications were published between 2001 and 2020, which followed an exponential growth pattern, with an annual growth rate of 11.6%. Radiology produced the most publications (n = 1290, 44.2%), followed by internal medicine (n = 716, 24.5%), surgery (n = 409, 14.0%), and otorhinolaryngology (n = 171, 5.9%). Otorhinolaryngology and internal medicine published significantly more case reports than radiology (p < 0.001, each). Radiology published a significantly higher proportion of publications on imaging diagnosis (p < 0.001 for all) and a significantly lower proportion of publications on biopsy (p < 0.001 for all) than the other disciplines. Publications produced by radiology authors were less frequently published in Q1 journals than those from other disciplines (p < 0.005 for internal medicine and miscellaneous disciplines and < 0.01 for surgery and otorhinolaryngology). China contributed the greatest number of publications (n = 622, 21.3%), followed by South Korea (n = 478, 16.4%) and the United States (n = 468, 16.0%).
Conclusion
Radiology produced the most publications for thyroid US than any other discipline. Radiology authors published more notably on imaging diagnosis compared to other topics and in journals with lower impact factors compared to authors in other disciplines.
8.Changes in Retinal Circulation Time after Panretinal Photocoagulation in Patients with Nonproliferative and Proliferative Diabetic Retinopathy
Ha Eun SIM ; Jeong Hyun LEE ; Jae Yong PARK ; Jin CHOI ; Min Ji KANG ; Je Hyung HWANG ; Jae Suk KIM
Journal of the Korean Ophthalmological Society 2021;62(4):507-515
Purpose:
To explore whether panretinal photocoagulation (PRP) improved retinal blood circulation in patients with diabetic retinopathy of various levels of severity based on changes in the retinal circulation time as measured by video fluorescein fundus angiography.
Methods:
We recruited patients with severe diabetic retinopathy indicated for PRP; we performed video fluorescein fundus angiography before and after PRP. We measured changes in the arterial circulation time (ACT), arteriovenous passage time (APT), and venous filling time (VFT) in patients with diabetic retinopathy of varying severity, and compared the findings. We recorded HbA1c levels and hypertension status.
Results:
We enrolled 33 patients, 17 with severe nonproliferative diabetic retinopathy (NPDR) and 16 with early proliferative diabetic retinopathy (PDR). After PRP, neither the ACT nor the APT changed significantly in patients with severe NPDR (p = 0.927, p = 0.138) or early PDR (p = 0.137, p = 0.268). After PRP, the VFT was significantly reduced in patients with both severe NPDR (p = 0.000) and early PDR (p = 0.022). The VFT reductions were similar in both groups (p = 0.217).
Conclusions
In patients with severe NPDR and early PDR, improvements in retinal blood circulation after PRP are reflected by reductions in the VFT evident on video fluorescein fundus angiography. We speculate that the improvements are similar in patients with severe NPDR and early PDR.
9.Changes in Retinal Circulation Time after Panretinal Photocoagulation in Patients with Nonproliferative and Proliferative Diabetic Retinopathy
Ha Eun SIM ; Jeong Hyun LEE ; Jae Yong PARK ; Jin CHOI ; Min Ji KANG ; Je Hyung HWANG ; Jae Suk KIM
Journal of the Korean Ophthalmological Society 2021;62(4):507-515
Purpose:
To explore whether panretinal photocoagulation (PRP) improved retinal blood circulation in patients with diabetic retinopathy of various levels of severity based on changes in the retinal circulation time as measured by video fluorescein fundus angiography.
Methods:
We recruited patients with severe diabetic retinopathy indicated for PRP; we performed video fluorescein fundus angiography before and after PRP. We measured changes in the arterial circulation time (ACT), arteriovenous passage time (APT), and venous filling time (VFT) in patients with diabetic retinopathy of varying severity, and compared the findings. We recorded HbA1c levels and hypertension status.
Results:
We enrolled 33 patients, 17 with severe nonproliferative diabetic retinopathy (NPDR) and 16 with early proliferative diabetic retinopathy (PDR). After PRP, neither the ACT nor the APT changed significantly in patients with severe NPDR (p = 0.927, p = 0.138) or early PDR (p = 0.137, p = 0.268). After PRP, the VFT was significantly reduced in patients with both severe NPDR (p = 0.000) and early PDR (p = 0.022). The VFT reductions were similar in both groups (p = 0.217).
Conclusions
In patients with severe NPDR and early PDR, improvements in retinal blood circulation after PRP are reflected by reductions in the VFT evident on video fluorescein fundus angiography. We speculate that the improvements are similar in patients with severe NPDR and early PDR.
10.2021 Korean Thyroid Imaging Reporting and Data System and Imaging-Based Management of Thyroid Nodules: Korean Society of Thyroid Radiology Consensus Statement and Recommendations
Eun Ju HA ; Sae Rom CHUNG ; Dong Gyu NA ; Hye Shin AHN ; Jin CHUNG ; Ji Ye LEE ; Jeong Seon PARK ; Roh-Eul YOO ; Jung Hwan BAEK ; Sun Mi BAEK ; Seong Whi CHO ; Yoon Jung CHOI ; Soo Yeon HAHN ; So Lyung JUNG ; Ji-hoon KIM ; Seul Kee KIM ; Soo Jin KIM ; Chang Yoon LEE ; Ho Kyu LEE ; Jeong Hyun LEE ; Young Hen LEE ; Hyun Kyung LIM ; Jung Hee SHIN ; Jung Suk SIM ; Jin Young SUNG ; Jung Hyun YOON ; Miyoung CHOI
Korean Journal of Radiology 2021;22(12):2094-2123
Incidental thyroid nodules are commonly detected on ultrasonography (US). This has contributed to the rapidly rising incidence of low-risk papillary thyroid carcinoma over the last 20 years. The appropriate diagnosis and management of these patients is based on the risk factors related to the patients as well as the thyroid nodules. The Korean Society of Thyroid Radiology (KSThR) published consensus recommendations for US-based management of thyroid nodules in 2011 and revised them in 2016. These guidelines have been used as the standard guidelines in Korea. However, recent advances in the diagnosis and management of thyroid nodules have necessitated the revision of the original recommendations. The task force of the KSThR has revised the Korean Thyroid Imaging Reporting and Data System and recommendations for US lexicon, biopsy criteria, US criteria of extrathyroidal extension, optimal thyroid computed tomography protocol, and US follow-up of thyroid nodules before and after biopsy. The biopsy criteria were revised to reduce unnecessary biopsies for benign nodules while maintaining an appropriate sensitivity for the detection of malignant tumors in small (1–2 cm) thyroid nodules. The goal of these recommendations is to provide the optimal scientific evidence and expert opinion consensus regarding US-based diagnosis and management of thyroid nodules.

Result Analysis
Print
Save
E-mail