HCHyeok Choi

Hyeok Choi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Hyeok Choi has a total of 31 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    A recurrent neural network-based predictive model for pressure ulcers based on an intensive care unit dataset (MIMIC-III) (Preprint)

    by Minseop Park, Hyeok Choi, Hee-Sung Ahn, Hee-Ju Kang, Saehoon Kim, Haeil Park, Jaeyong Shin, Soo Ick Cho

    BACKGROUND A pressure ulcer (PU) is a localized cutaneous injury caused by pressure or shear, which usually occurs in the region of a bony prominence. PUs are common in hospitalized patients and cause complications including infection. OBJECTIVE This study aimed to build a recurrent neural network-based algorithm to predict PUs 24 hours before their occurrence. METHODS This study analyzed a freely accessible intensive care unit (ICU) dataset, MIMIC- III. Deep learning and machine learning algorithms including long short-term memory (LSTM), multilayer perceptron (MLP), and XGBoost were applied to 37 dynamic features (including the Braden scale, vital signs and laboratory results, and interventions to reduce the risk of PUs) and 35 static features (including the length of time spent in the ICU, demographics, and comorbidities). Their outcomes were compared in terms of the area under the receiver operating characteristic (AUROC) and the area under the precision-recall curve (AUPRC). RESULTS A total of 1,048 cases of PUs (10.0%) and 9,402 controls (90.0%) without PUs satisfied the inclusion criteria for analysis. The LSTM + MLP model (AUROC: 0.7929 ± 0.0095, AUPRC: 0.4819 ± 0.0109) outperformed the other models, namely: MLP model (AUROC: 0.7777 ± 0.0083, AUPRC: 0.4527 ± 0.0195) and XGBoost (AUROC: 0.7465 ± 0.0087, AUPRC: 0.4052 ± 0.0087). Various features, including the length of time spent in the ICU, Glasgow coma scale, and the Braden scale, contributed to the prediction model. CONCLUSIONS This study suggests that recurrent neural network-based algorithms such as LSTM can be applied to evaluate the risk of PUs in ICU patients.

  • Unclassifiable repeated antineutrophil cytoplasmic antibody (ANCA) positivity in diseases other than ANCA-associated vasculitis

    by Hyeok Choi, Yong-Beom Park, Jaewoo Song, Sang-Won Lee

    Background and objective Antineutrophil cytoplasmic antibody (ANCA) is a specific autoantibody for ANCA-associated vasculitis (AAV). However, ANCA can be detected in various diseases other than AAV. Hence, in this study, we investigated and provided the name of diseases with repeated ANCA positivity and the frequency of each disease other than AAV. Methods We retrospectively screened the results of the tests of ANCA in 26,499 patients using the Clinical Data Repository System and included in this study only 173 patients with repeated ANCA positivity more than twice. ‘Unclassifiable ANCA’ was defined when ANCA was detected in patients with diseases other than AAV. ‘Unclassifiable repeated ANCA’ was also defined when unclassifiable ANCA was successively detected more than twice. Results Among rheumatic and autoimmune diseases, the most common disease with unclassifiable repeated ANCA was vasculitis undetermined (21.0%). In terms of cardiovascular and cerebrovascular diseases, the most common disease with unclassifiable repeated ANCA was atherosclerotic heart disease (12.1%). In terms of disorders in liver, kidneys and lungs, the most common disease with unclassifiable repeated ANCA was chronic kidney disease (51 cases, 29.5%). In addition, among infections with confirmed infectious pathogens, the most common pathogen with unclassifiable repeated ANCA was varicella-zoster virus (6.9%) followed by Candida (4.6%). Conclusion Overall, regardless of category, the common diseases with unclassifiable repeated ANCA were chronic kidney disease followed by interstitial lung disease and vasculitis undetermined. Thus, we carefully suggest that physicians should pay more attention to the development of AAV or vasculitis other than AAV and, furthermore, kidneys and lungs should be monitored regularly and closely in patients with unclassifiable repeated ANCA.


  • Can antineutrophil cytoplasmic antibody positivity at diagnosis predict the poor outcomes of Sjögren’s syndrome?

    by Soo Bin Lee, Hyeok Choi, Minyoung Kevin Kim, Seung Min Jung, Jason Jungsik Song, Yong-Beom Park, Sang-Won Lee

    We investigated the clinical implication of ANCA positivity at diagnosis on the poor outcomes in patients with Sjögren’s syndrome. The medical records of 606 Korean patients with Sjögren’s syndrome were retrospectively reviewed. The results of perinuclear (P)-ANCA, myeloperoxidase (MPO)-ANCA, cytoplasmic (C)-ANCA, and proteinase 3 (PR3)-ANCA were collected and the frequencies of all-cause mortality, interstitial lung disease (ILD), end-stage renal disease (ESRD), and lymphoma were assessed as the poor outcomes of Sjögren’s syndrome. Comparison of the cumulative patient survivals between the two groups was analysed by the Kaplan–Meier survival analysis. Of the 606 patients, ANCA was detected in 10.2% of Sjögren’s syndrome patients without AAV. Twenty-one patients (3.5%) died, 99 patients (16.3%) suffered from ILD, and 8 patients had ESRD. Lymphoma occurred in 5 patients (0.8%) during 37.5 months. Sjögren’s syndrome patients with ANCA positivity exhibited a lower cumulative ILD-free survival rate than those with ANCA negativity (P = 0.001). Sjögren’s syndrome patients with P-ANCA positivity and those with MPO-ANCA (or P-ANCA) positivity showed a lower cumulative ILD-free survival rate than those without (P = 0.012 and P < 0.001). Also, Sjögren’s syndrome patients with P-ANCA positivity exhibited a lower cumulative ESRD-free survival rate than those without (P = 0.043). ANCA positivity was associated with neither all-cause mortality nor lymphoma in Sjögren’s syndrome patients. ANCA positivity and MPO-ANCA (or P-ANCA) positivity at diagnosis was associated with the development of ILD during follow-up in patients with Sjögren’s syndrome.


  • Open Access

    Multivariable index for assessing the activity and predicting all‐cause mortality in antineutrophil cytoplasmic antibody‐associated vasculitis

    by Minyoung Kevin Kim, Hyeok Choi, Jae Yeon Kim, Jason Jungsik Song, Yong‐Beom Park, Sang‐Won Lee

    Background: So far, there has been no tool to estimate activity at diagnosis and predict all‐cause mortality in patients with ANCA‐associated vasculitis (AAV). Hence, we determined the initial predictors of them in patients with AAV. Methods: We retrospectively reviewed the medical records of 182 patients with AAV. Severe AAV was defined as Birmingham Vasculitis Activity Score (BVAS) ≥ 16. The cutoffs were extrapolated by the receiver operator characteristic (ROC) curve. The odds ratio (OR) and the relative risk (RR) were assessed using the multivariable logistic regression analysis and the chi‐square test, respectively. Results: In the comparison analysis, patients with severe AAV exhibited the higher neutrophil and platelet counts, creatinine, erythrocyte sedimentation rate and C‐reactive protein, and the lower lymphocyte count, hemoglobin, and serum albumin than those without. In the multivariable logistic regression analysis, creatinine ≥ 0.9 mg/dL (OR 2.264), lymphocyte count ≤ 1430.0/mm 3 (OR 1.856), and hemoglobin ≤ 10.8 g/dL (OR 2.085) were associated with severe AAV. We developed a new equation of a multivariable index for AAV (MVIA) = 0.6 × (Lymphocyte count ≤ 1430.0/mm 3 ) + 0.7 × (Hemoglobin ≤ 10.8 g/dL) + 0.8 × (Creatinine ≥ 0.9 mg/dL). The optimal cutoff of MVIA for severe AAV was set as 1.35. Severe AAV was identified more frequently in patients with MVIA at diagnosis ≥1.35 than those without (RR 4.432). Patients with MVIA at diagnosis ≥1.35 exhibited the lower cumulative patient survival rate than those without. Conclusion: Multivariable index for AAV could assess the cross‐sectional activity and predict all‐cause mortality in patients with AAV.


  • Clinical significance of ANCA positivity in patients with IgA vasculitis: a retrospective monocentric study

    by Jae Yeon Kim, Hyeok Choi, Minyoung Kevin Kim, Soo Bin Lee, Yong-Beom Park, Sang-Won Lee

    We assessed the detection rate of antineutrophil cytoplasmic antibody (ANCA) and investigated the clinical significance of ANCA positivity at diagnosis in patients with IgA vasculitis (Henoch-Schönlein purpura). We retrospectively reviewed their medical records of 86 IgA vasculitis patients. We divided IgA vasculitis patients based on ANCA positivity and compared variables at diagnosis and poor outcomes and medication during follow-up between the two groups. All-cause mortality, relapse, chronic kidney disease (CKD) (stage 3–5) and end-stage renal disease (ESRD) were defined as poor outcomes. We assessed the renal histological features based on the International Study of Kidney Disease in Children (ISKDC) classification and Oxford classification. Comparison of cumulative survivals was analysed by the Kaplan–Meier survival analysis. Five of 86 IgA vasculitis patients (5.8%) had ANCA and all ANCA-positive patients had myeloperoxidase (MPO)-ANCA. IgA vasculitis patients with ANCA exhibited pulmonary and nervous involvement of IgA vasculitis more frequently than those without. There was no significant difference in renal involvement between the two groups. There were no significant differences in renal histological features and poor outcomes related to renal function between IgA vasculitis patients with and without ANCA. In addition, 5 IgA vasculitis patients did not meet the classification criteria for ANCA-associated vasculitis (AAV). Particularly, there were no significant differences in CKD and ESRD-free survival rates between IgA vasculitis patients with and without ANCA. 5.8% of IgA vasculitis patients had MPO-ANCA and poor outcomes of IgA vasculitis were not affected by the presence of ANCA.


  • Anti-phospholipid antibody syndrome occurrence in patients with persistent anti-phospholipid antibodies

    by Hyeok Choi, Sung Soo Ahn, Jason Jungsik Song, Yong-Beom Park, Jaewoo Song, Sang-Won Lee

    We investigated the overall frequency of anti-phospholipid syndrome (APS) occurrence in Korean patients with consecutively detected anti-phospholipid antibodies with an interval of 12 weeks (persistent aPLs). We retrospectively reviewed the results of blood tests of aPLs in 14,889 patients in whom aPL tests were performed at Yonsei University College of Medicine, Severance Hospital, from January 2012 to August 2018, and included 833 patients with persistent aPLs. We obtained clinical and laboratory data including anti-cardiolipin antibodies IgM and IgG, anti-beta2 glycoprotein1 IgM and IgG, and lupus anticoagulant (LAC). Of 833 patients with persistent aPLs, 96 patients (11.5%) had APS (84 patients had thrombotic events and 12 had pregnancy morbidity). Among aPLs, LAC was detected in patients with APS more frequently than asymptomatic carriers of aPLs (46.9% vs. 25.9%, p < 0.001). Patients with LAC (relative risk (RR) 2.558, p < 0.001) and aPLs ≥ 2 (RR 1.731, p = 0.014) exhibited the higher rate of APS occurrence than those without. Moreover, patients with aPLs ≥ 3 and aPLs ≥ 4 exhibited the higher rates of APS occurrence than those without (RR 2.753, p < 0.001 and RR 3.209, p = 0.013). Meanwhile, patients with ANA, anti-dsDNA, anti-SSA/Ro, and SLE exhibited the increased frequency of LAC positivity, compared to those without (RR 3.304, p = 0.005, RR 4.269, p = 0.032, RR 3.750, p = 0.041 and RR 8.828, p < 0.001, respectively). APS occurs in 11.5% of Korean patients with persistent aPLs. LAC positivity and aPLs ≥ 2 were significantly associated with APS occurrence. SLE and SLE-related autoantibodies were associated with LAC positivity.


  • Systemic immune‐inflammation index could estimate the cross‐sectional high activity and the poor outcomes in immunosuppressive drug‐naïve patients with antineutrophil cytoplasmic antibody‐associated vasculitis

    by Youhyun Kim, Hyeok Choi, Seung Min Jung, Jason Jungsik Song, Yong‐Beom Park, Sang‐Won Lee

    Objectives: We investigated whether systemic immune‐inflammation index (SII) at diagnosis can estimate the cross‐sectional high activity and predict the poor outcomes in immunosuppressive drug‐naïve patients with antineutrophil cytoplasmic antibody‐associated vasculitis (AAV).Methods: We retrospectively reviewed the medical records of 163 patients with AAV and obtained clinical and laboratory data. We calculated Birmingham vasculitis activity score (BVAS) as well as five‐factor score (FFS) (2009) at diagnosis. SII at diagnosis was calculated by the equation of (SII at diagnosis = platelet count × neutrophil count/lymphocyte count at diagnosis). Severe AAV was defined as BVAS at diagnosis ≥16. The odds ratio was assessed using the multivariable logistic regression analysis and cumulative survival rates were compared by the Kaplan–Meier survival analysis.Results: The median age at diagnosis was 58.0 years old and 51 patients were men. The median BVAS was 12.0. Fifty‐seven patients had severe AAV. The median SII at diagnosis was 1349.6. In the multivariable analysis, only SII exhibited a significant odds ratio for the cross‐sectional severe AAV (P = 0.043). We obtained the cut‐off of SII at diagnosis for severe AAV as 1573.56. Patients with SII at diagnosis ≥1573.56 exhibited a significantly high relative risk of the cross‐sectional severe AAV compared to those without (relative risk 4.625). Furthermore, patients with SII at diagnosis ≥1573.56 exhibited significantly the lower cumulative relapse free and renal survivals than those without.Conclusion: Systemic immune‐inflammation index at diagnosis could estimate the cross‐section severe AAV and predict the poor outcomes in AAV patients.


  • Low serum complement 3 level is associated with severe ANCA-associated vasculitis at diagnosis

    by Hyeok Choi, Youhyun Kim, Seung Min Jung, Jason Jungsik Song, Yong-Beom Park, Sang-Won Lee

    We investigated whether low serum C3 level can cross-sectionally estimate severe antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) in immunosuppressive drug-naïve patients at diagnosis.