Se A. Lee
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Se A. Lee has a total of 54 co-authors. Here are the top co-authors by number of shared publications:
- Jong Dae Lee
Department of Otorhinolaryngology-Head and Neck Surgery, Soonchunhyang University Bucheon Hospital, Soonchunhyang University School of Medicine, Bucheon, Korea
7 shared publications - Jae Young Choi
Department of Otorhinolaryngology, Yonsei University College of Medicine, Seoul 03722, Republic of Korea
3 shared publications - Ara Cho
Yonsei Univ, Dept Otorhinolaryngol, Coll Med, 50-1 Yonsei Ro, Seoul 03722, South Korea
2 shared publications - Bo Gyung Kim
Department of Otorhinolaryngology-Head and Neck Surgery, Soonchunhyang University College of Medicine, Bucheon, Korea
2 shared publications - Chongwon Pae
Department of Psychiatry, CHA Bundang Medical Center, CHA University School of Medicine, Seongnam, Republic of Korea
2 shared publications - Dongchul Cha
Healthcare Lab, Naver Corporation, Seongnam, Korea
2 shared publications
Latest Publications
- Open Access
- 1 January 2026
- Published by XMLink in Journal of Korean Medical Science
- Vol. 41 (24) , e166
Values and Preferences of Korean Patients With Facial Palsy: A Cross-Sectional Survey
Joo HA, et al. J Korean Med Sci. 2026 Jun;41(24):e166. https://doi.org/10.3346/jkms.2026.41.e166
- 1 January 2026
- Published by XMLink in Journal of Korean Medical Science
- Vol. 41 (24) , e166
Open AccessValues and Preferences of Korean Patients With Facial Palsy: A Cross-Sectional Survey
Joo HA, et al. J Korean Med Sci. 2026 Jun;41(24):e166. https://doi.org/10.3346/jkms.2026.41.e166
- Open Access
- 3 December 2025
- Published by Elsevier in Jornal de Pediatria
- Vol. 102 (1) , 101479
Respiratory mechanic instability in evaluating the effectiveness of adenotonsillectomy for pediatric obstructive sleep apnea
- 3 December 2025
- Published by Elsevier in Jornal de Pediatria
- Vol. 102 (1) , 101479
Open AccessRespiratory mechanic instability in evaluating the effectiveness of adenotonsillectomy for pediatric obstructive sleep apnea
- Open Access
- 1 January 2024
- Published by XMLink in Journal of Korean Medical Science
- Vol. 39 (16) , e140
The Impact of the COVID-19 Pandemic on Bell’s Palsy and Ramsay-Hunt Syndrome: A Multicenter Retrospective Study
Kwak MY, et al. J Korean Med Sci. 2024 Apr;39(16):e140. https://doi.org/10.3346/jkms.2024.39.e140
- 1 January 2024
- Published by XMLink in Journal of Korean Medical Science
- Vol. 39 (16) , e140
Open AccessThe Impact of the COVID-19 Pandemic on Bell’s Palsy and Ramsay-Hunt Syndrome: A Multicenter Retrospective Study
Kwak MY, et al. J Korean Med Sci. 2024 Apr;39(16):e140. https://doi.org/10.3346/jkms.2024.39.e140
- Open Access
- 30 November 2023
- Published by Korean Society of Otorhinolaryngology-Head and Neck Surgery in Clinical and Experimental Otorhinolaryngology
- Vol. 16 (4) , 403-406
Genetic Characteristics and Audiological Performance After Cochlear Implantation in Patients With Incomplete Partition Type III
- 30 November 2023
- Published by Korean Society of Otorhinolaryngology-Head and Neck Surgery in Clinical and Experimental Otorhinolaryngology
- Vol. 16 (4) , 403-406
Open AccessGenetic Characteristics and Audiological Performance After Cochlear Implantation in Patients With Incomplete Partition Type III
- Open Access
- 1 July 2023
- Published by American Medical Association (AMA) in JAMA Otolaryngology–Head & Neck Surgery
- Vol. 149 (7) , 571-578
Association of Hearing Loss With Anatomical and Functional Connectivity in Patients With Mild Cognitive Impairment
Alzheimer disease (AD) is characterized by the deposition of β-amyloid into plaques and hyperphosphorylated τ as neurofibrillary tangles, which have been shown to be signficantly associated with neuronal dysfunction and subsequent cognitive impairment.1 Better AD prevention and/or postponement may be possible with a better understanding of the mechanistic underpinnings of why a sizeable portion of older adults maintain a normal cognitive function despite AD pathologies.2 Less education, hypertension, hearing loss, smoking, obesity, depression, physical inactivity, diabetes, low social contact, excessive alcohol consumption, traumatic brain injury, and air pollution have been shown to be modifiable risk factors for AD.3 Studies have also suggested that hearing loss is an important modifiable risk factor for AD.4-8
- 1 July 2023
- Published by American Medical Association (AMA) in JAMA Otolaryngology–Head & Neck Surgery
- Vol. 149 (7) , 571-578
Open AccessAssociation of Hearing Loss With Anatomical and Functional Connectivity in Patients With Mild Cognitive Impairment
Alzheimer disease (AD) is characterized by the deposition of β-amyloid into plaques and hyperphosphorylated τ as neurofibrillary tangles, which have been shown to be signficantly associated with neuronal dysfunction and subsequent cognitive impairment.1 Better AD prevention and/or postponement may be possible with a better understanding of the mechanistic underpinnings of why a sizeable portion of older adults maintain a normal cognitive function despite AD pathologies.2 Less education, hypertension, hearing loss, smoking, obesity, depression, physical inactivity, diabetes, low social contact, excessive alcohol consumption, traumatic brain injury, and air pollution have been shown to be modifiable risk factors for AD.3 Studies have also suggested that hearing loss is an important modifiable risk factor for AD.4-8
- Open Access
- 30 April 2023
- Published by The Korean Academy of Geriatrics & Gerontology in Korean Journal of Geriatrics & Gerontology
- Vol. 24 (1) , 12-16
Approaches to Age-Related Hearing Loss
- 30 April 2023
- Published by The Korean Academy of Geriatrics & Gerontology in Korean Journal of Geriatrics & Gerontology
- Vol. 24 (1) , 12-16
Open AccessApproaches to Age-Related Hearing Loss
- Open Access
- 31 December 2022
- Published by The Busan, Ulsan, Gyeoungnam Branch of The Korean Society of Otolaryngology in Journal of Clinical Otolaryngology
- Vol. 33 (4) , 165-173
Pathophysiology and Classification of Eustachian Tube Disorder
임상이비인후과(Journal of Clinical Otolaryngology Head and Neck Surgery)
- 31 December 2022
- Published by The Busan, Ulsan, Gyeoungnam Branch of The Korean Society of Otolaryngology in Journal of Clinical Otolaryngology
- Vol. 33 (4) , 165-173
Open AccessPathophysiology and Classification of Eustachian Tube Disorder
임상이비인후과(Journal of Clinical Otolaryngology Head and Neck Surgery)
- Open Access
- 21 November 2022
- Published by Korean Society of Otorhinolaryngology-Head and Neck Surgery in Korean Journal of Otorhinolaryngology-Head and Neck Surgery
- Vol. 65 (11) , 678-683
Functional Results and Learning Curve of Endoscopic Ossiculoplasty
Background and Objectives Endoscopes are recently being used to manage various middle ear surgeries including ossiculoplasty. Here, we investigated hearing results of endoscopic ossiculoplasty and surgeon’ learning curve in gaining endoscopic ossiculoplasty proficiency. We compared the results with those of microscopic ossiculoplasty.Subjects and Method We analyzed 48 patients of ossiculoplasties performed from 2010 to 2020. Patients were divided into three groups: microscopic technique group (n=16); endoscopic early phase group (n=16); endoscopic late phase group (n=16). Operation time and hearing results including postoperative air-bone gap (ABG) and ABG closure were analyzed. Hearing outcomes were considered “successful” when the postoperative ABG was less than 20 dB and considered “excellent” when ABG was less than 10 dB. The learning curve of the outcomes were analyzed.Results Both operation time and postoperative ABG showed decreasing tendency, but only operation time showed statistical significance. The ratio of successful and excellent operations improved, but did not show statistical significance. Improvement in learning curve was not statistically significant presumably due to the lack of patient numbers.Conclusion Surgeon’s proficiency of endoscopic ossiculoplasty was correlated with the improved clinical outcomes, and proficiently conducted endoscopic ossiculoplasty could be used as a surgical technique comparable to microscopic ossiculoplasty.
- 21 November 2022
- Published by Korean Society of Otorhinolaryngology-Head and Neck Surgery in Korean Journal of Otorhinolaryngology-Head and Neck Surgery
- Vol. 65 (11) , 678-683
Open AccessFunctional Results and Learning Curve of Endoscopic Ossiculoplasty
Background and Objectives Endoscopes are recently being used to manage various middle ear surgeries including ossiculoplasty. Here, we investigated hearing results of endoscopic ossiculoplasty and surgeon’ learning curve in gaining endoscopic ossiculoplasty proficiency. We compared the results with those of microscopic ossiculoplasty.Subjects and Method We analyzed 48 patients of ossiculoplasties performed from 2010 to 2020. Patients were divided into three groups: microscopic technique group (n=16); endoscopic early phase group (n=16); endoscopic late phase group (n=16). Operation time and hearing results including postoperative air-bone gap (ABG) and ABG closure were analyzed. Hearing outcomes were considered “successful” when the postoperative ABG was less than 20 dB and considered “excellent” when ABG was less than 10 dB. The learning curve of the outcomes were analyzed.Results Both operation time and postoperative ABG showed decreasing tendency, but only operation time showed statistical significance. The ratio of successful and excellent operations improved, but did not show statistical significance. Improvement in learning curve was not statistically significant presumably due to the lack of patient numbers.Conclusion Surgeon’s proficiency of endoscopic ossiculoplasty was correlated with the improved clinical outcomes, and proficiently conducted endoscopic ossiculoplasty could be used as a surgical technique comparable to microscopic ossiculoplasty.
- Open Access
- 8 December 2021
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 9 (12) , e33049
Differential Biases and Variabilities of Deep Learning–Based Artificial Intelligence and Human Experts in Clinical Diagnosis: Retrospective Cohort and Survey Study
Background: Deep learning (DL)–based artificial intelligence may have different diagnostic characteristics than human experts in medical diagnosis. As a data-driven knowledge system, heterogeneous population incidence in the clinical world is considered to cause more bias to DL than clinicians. Conversely, by experiencing limited numbers of cases, human experts may exhibit large interindividual variability. Thus, understanding how the 2 groups classify given data differently is an essential step for the cooperative usage of DL in clinical application.Objective: This study aimed to evaluate and compare the differential effects of clinical experience in otoendoscopic image diagnosis in both computers and physicians exemplified by the class imbalance problem and guide clinicians when utilizing decision support systems.Methods: We used digital otoendoscopic images of patients who visited the outpatient clinic in the Department of Otorhinolaryngology at Severance Hospital, Seoul, South Korea, from January 2013 to June 2019, for a total of 22,707 otoendoscopic images. We excluded similar images, and 7500 otoendoscopic images were selected for labeling. We built a DL-based image classification model to classify the given image into 6 disease categories. Two test sets of 300 images were populated: balanced and imbalanced test sets. We included 14 clinicians (otolaryngologists and nonotolaryngology specialists including general practitioners) and 13 DL-based models. We used accuracy (overall and per-class) and kappa statistics to compare the results of individual physicians and the ML models.Results: Our ML models had consistently high accuracies (balanced test set: mean 77.14%, SD 1.83%; imbalanced test set: mean 82.03%, SD 3.06%), equivalent to those of otolaryngologists (balanced: mean 71.17%, SD 3.37%; imbalanced: mean 72.84%, SD 6.41%) and far better than those of nonotolaryngologists (balanced: mean 45.63%, SD 7.89%; imbalanced: mean 44.08%, SD 15.83%). However, ML models suffered from class imbalance problems (balanced test set: mean 77.14%, SD 1.83%; imbalanced test set: mean 82.03%, SD 3.06%). This was mitigated by data augmentation, particularly for low incidence classes, but rare disease classes still had low per-class accuracies. Human physicians, despite being less affected by prevalence, showed high interphysician variability (ML models: kappa=0.83, SD 0.02; otolaryngologists: kappa=0.60, SD 0.07).Conclusions: Even though ML models deliver excellent performance in classifying ear disease, physicians and ML models have their own strengths. ML models have consistent and high accuracy while considering only the given image and show bias toward prevalence, whereas human physicians have varying performance but do not show bias toward prevalence and may also consider extra information that is not images. To deliver the best patient care in the shortage of otolaryngologists, our ML model can serve a cooperative role for clinicians with diverse expertise, as long as it is kept in mind that models consider only images and could be biased toward prevalent diseases even after data augmentation.
- 8 December 2021
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 9 (12) , e33049
Open AccessDifferential Biases and Variabilities of Deep Learning–Based Artificial Intelligence and Human Experts in Clinical Diagnosis: Retrospective Cohort and Survey Study
Background: Deep learning (DL)–based artificial intelligence may have different diagnostic characteristics than human experts in medical diagnosis. As a data-driven knowledge system, heterogeneous population incidence in the clinical world is considered to cause more bias to DL than clinicians. Conversely, by experiencing limited numbers of cases, human experts may exhibit large interindividual variability. Thus, understanding how the 2 groups classify given data differently is an essential step for the cooperative usage of DL in clinical application.Objective: This study aimed to evaluate and compare the differential effects of clinical experience in otoendoscopic image diagnosis in both computers and physicians exemplified by the class imbalance problem and guide clinicians when utilizing decision support systems.Methods: We used digital otoendoscopic images of patients who visited the outpatient clinic in the Department of Otorhinolaryngology at Severance Hospital, Seoul, South Korea, from January 2013 to June 2019, for a total of 22,707 otoendoscopic images. We excluded similar images, and 7500 otoendoscopic images were selected for labeling. We built a DL-based image classification model to classify the given image into 6 disease categories. Two test sets of 300 images were populated: balanced and imbalanced test sets. We included 14 clinicians (otolaryngologists and nonotolaryngology specialists including general practitioners) and 13 DL-based models. We used accuracy (overall and per-class) and kappa statistics to compare the results of individual physicians and the ML models.Results: Our ML models had consistently high accuracies (balanced test set: mean 77.14%, SD 1.83%; imbalanced test set: mean 82.03%, SD 3.06%), equivalent to those of otolaryngologists (balanced: mean 71.17%, SD 3.37%; imbalanced: mean 72.84%, SD 6.41%) and far better than those of nonotolaryngologists (balanced: mean 45.63%, SD 7.89%; imbalanced: mean 44.08%, SD 15.83%). However, ML models suffered from class imbalance problems (balanced test set: mean 77.14%, SD 1.83%; imbalanced test set: mean 82.03%, SD 3.06%). This was mitigated by data augmentation, particularly for low incidence classes, but rare disease classes still had low per-class accuracies. Human physicians, despite being less affected by prevalence, showed high interphysician variability (ML models: kappa=0.83, SD 0.02; otolaryngologists: kappa=0.60, SD 0.07).Conclusions: Even though ML models deliver excellent performance in classifying ear disease, physicians and ML models have their own strengths. ML models have consistent and high accuracy while considering only the given image and show bias toward prevalence, whereas human physicians have varying performance but do not show bias toward prevalence and may also consider extra information that is not images. To deliver the best patient care in the shortage of otolaryngologists, our ML model can serve a cooperative role for clinicians with diverse expertise, as long as it is kept in mind that models consider only images and could be biased toward prevalent diseases even after data augmentation.
- Open Access
- 22 August 2021
- Published by JMIR Publications Inc. in JMIR Preprints
Differential Biases and Variabilities of Deep Learning–Based Artificial Intelligence and Human Experts in Clinical Diagnosis: Retrospective Cohort and Survey Study (Preprint)
BACKGROUND Deep learning (DL)–based artificial intelligence may have different diagnostic characteristics than human experts in medical diagnosis. As a data-driven knowledge system, heterogeneous population incidence in the clinical world is considered to cause more bias to DL than clinicians. Conversely, by experiencing limited numbers of cases, human experts may exhibit large interindividual variability. Thus, understanding how the 2 groups classify given data differently is an essential step for the cooperative usage of DL in clinical application. OBJECTIVE This study aimed to evaluate and compare the differential effects of clinical experience in otoendoscopic image diagnosis in both computers and physicians exemplified by the class imbalance problem and guide clinicians when utilizing decision support systems. METHODS We used digital otoendoscopic images of patients who visited the outpatient clinic in the Department of Otorhinolaryngology at Severance Hospital, Seoul, South Korea, from January 2013 to June 2019, for a total of 22,707 otoendoscopic images. We excluded similar images, and 7500 otoendoscopic images were selected for labeling. We built a DL-based image classification model to classify the given image into 6 disease categories. Two test sets of 300 images were populated: balanced and imbalanced test sets. We included 14 clinicians (otolaryngologists and nonotolaryngology specialists including general practitioners) and 13 DL-based models. We used accuracy (overall and per-class) and kappa statistics to compare the results of individual physicians and the ML models. RESULTS Our ML models had consistently high accuracies (balanced test set: mean 77.14%, SD 1.83%; imbalanced test set: mean 82.03%, SD 3.06%), equivalent to those of otolaryngologists (balanced: mean 71.17%, SD 3.37%; imbalanced: mean 72.84%, SD 6.41%) and far better than those of nonotolaryngologists (balanced: mean 45.63%, SD 7.89%; imbalanced: mean 44.08%, SD 15.83%). However, ML models suffered from class imbalance problems (balanced test set: mean 77.14%, SD 1.83%; imbalanced test set: mean 82.03%, SD 3.06%). This was mitigated by data augmentation, particularly for low incidence classes, but rare disease classes still had low per-class accuracies. Human physicians, despite being less affected by prevalence, showed high interphysician variability (ML models: kappa=0.83, SD 0.02; otolaryngologists: kappa=0.60, SD 0.07). CONCLUSIONS Even though ML models deliver excellent performance in classifying ear disease, physicians and ML models have their own strengths. ML models have consistent and high accuracy while considering only the given image and show bias toward prevalence, whereas human physicians have varying performance but do not show bias toward prevalence and may also consider extra information that is not images. To deliver the best patient care in the shortage of otolaryngologists, our ML model can serve a cooperative role for clinicians with diverse expertise, as long as it is kept in mind that models consider only images and could be biased toward prevalent diseases even after data augmentation.
- 22 August 2021
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessDifferential Biases and Variabilities of Deep Learning–Based Artificial Intelligence and Human Experts in Clinical Diagnosis: Retrospective Cohort and Survey Study (Preprint)
BACKGROUND Deep learning (DL)–based artificial intelligence may have different diagnostic characteristics than human experts in medical diagnosis. As a data-driven knowledge system, heterogeneous population incidence in the clinical world is considered to cause more bias to DL than clinicians. Conversely, by experiencing limited numbers of cases, human experts may exhibit large interindividual variability. Thus, understanding how the 2 groups classify given data differently is an essential step for the cooperative usage of DL in clinical application. OBJECTIVE This study aimed to evaluate and compare the differential effects of clinical experience in otoendoscopic image diagnosis in both computers and physicians exemplified by the class imbalance problem and guide clinicians when utilizing decision support systems. METHODS We used digital otoendoscopic images of patients who visited the outpatient clinic in the Department of Otorhinolaryngology at Severance Hospital, Seoul, South Korea, from January 2013 to June 2019, for a total of 22,707 otoendoscopic images. We excluded similar images, and 7500 otoendoscopic images were selected for labeling. We built a DL-based image classification model to classify the given image into 6 disease categories. Two test sets of 300 images were populated: balanced and imbalanced test sets. We included 14 clinicians (otolaryngologists and nonotolaryngology specialists including general practitioners) and 13 DL-based models. We used accuracy (overall and per-class) and kappa statistics to compare the results of individual physicians and the ML models. RESULTS Our ML models had consistently high accuracies (balanced test set: mean 77.14%, SD 1.83%; imbalanced test set: mean 82.03%, SD 3.06%), equivalent to those of otolaryngologists (balanced: mean 71.17%, SD 3.37%; imbalanced: mean 72.84%, SD 6.41%) and far better than those of nonotolaryngologists (balanced: mean 45.63%, SD 7.89%; imbalanced: mean 44.08%, SD 15.83%). However, ML models suffered from class imbalance problems (balanced test set: mean 77.14%, SD 1.83%; imbalanced test set: mean 82.03%, SD 3.06%). This was mitigated by data augmentation, particularly for low incidence classes, but rare disease classes still had low per-class accuracies. Human physicians, despite being less affected by prevalence, showed high interphysician variability (ML models: kappa=0.83, SD 0.02; otolaryngologists: kappa=0.60, SD 0.07). CONCLUSIONS Even though ML models deliver excellent performance in classifying ear disease, physicians and ML models have their own strengths. ML models have consistent and high accuracy while considering only the given image and show bias toward prevalence, whereas human physicians have varying performance but do not show bias toward prevalence and may also consider extra information that is not images. To deliver the best patient care in the shortage of otolaryngologists, our ML model can serve a cooperative role for clinicians with diverse expertise, as long as it is kept in mind that models consider only images and could be biased toward prevalent diseases even after data augmentation.