YCYoungjin Cho

Youngjin Cho

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Youngjin Cho has a total of 232 co-authors. Here are the top co-authors by number of shared publications:

  • Joonghee Kim

    Department of Emergency Medicine, Seoul National University Bundang Hospital, Seongnam-si, Gyeonggi-do, Republic of Korea

    23 shared publications
  • Il-Young Oh

    Cardiovascular Center, Seoul National University Bundang Hospital, Seoul National University College of Medicine, Seongnam, Korea

    15 shared publications
  • Ji Hyun Lee

    Department of Cardiology, Cardiovascular Center, Seoul National University Bundang Hospital, Seongnam, Korea

    13 shared publications
  • Il-Young Oh

    Division of Cardiology, Department of Internal Medicine, Seoul National University Bundang Hospital, Seongnam, Republic of Korea

    9 shared publications
  • Ji Hyun Lee

    Division of Cardiology, Department of Internal Medicine, Seoul National University Bundang Hospital, Seongnam, Republic of Korea

    8 shared publications
  • Joonghee Kim

    Department of Emergency Medicine, Seoul National University Bundang Hospital, Seongnam, Republic of Korea

    8 shared publications

Latest Publications

  • Open Access

    Multitask Artificial Intelligence–Based Electrocardiogram Tool for Preoperative Cardiac Testing in Noncardiac Surgery: Retrospective Cohort Study of Health Care Utilization and Costs

    by Hong-Mi Choi, Yerin Kim, Joonghee Kim, Jiesuck Park, In-Chang Hwang, Yun Young Choi, Ji Hyun Lee, Yeonyee E Yoon, Il-Young Oh, Goo-Yeong Cho, et al.

    Background: Preoperative cardiovascular risk stratification is essential in noncardiac surgery, but conventional testing is frequently overused, increasing costs without improving outcomes. Artificial intelligence (AI)–enabled electrocardiography (ECG) may enhance perioperative risk assessment by identifying surgical candidates at very low-risk for adverse events. Objective: This study aimed to evaluate whether AI-ECG–based risk stratification could help decrease low-yield preoperative cardiovascular testing and reduce associated costs, without an observed increase in postoperative adverse outcomes, in candidates for noncardiac surgery. Methods: We retrospectively analyzed 41,218 patients (46,135 ECG-surgery pairs) undergoing noncardiac surgery at Seoul National University Bundang Hospital (2020-2021). An AI-ECG algorithm generated eight probability scores for cardiac conditions, classifying cases as low- or high-risk. Based on the performance and results of preoperative cardiovascular testing (transthoracic echocardiography, coronary computed tomography angiography, single-photon emission computed tomography, or coronary angiography), cases were classified as no advanced cardiovascular imaging, negative-test, or positive-test. The primary end point was a 30-day composite of all-cause mortality and unplanned percutaneous coronary intervention. Results: AI-ECG classified 92.3% (42,599/46,135) of cases as low-risk, with a composite outcome rate of 0.2% (79/42,599) vs 2.9% (101/3536) in high-risk cases. Preoperative cardiovascular testing was performed in 11.8% (5458/46,135) of cases, with only 16.3% (892/5458) yielding positive findings. In AI-ECG low-risk cases, event rates were uniformly low (0.2%-0.4%) irrespective of whether advanced cardiovascular testing was performed, whereas in high-risk cases, rates were consistently high (2.6%-3.4%). The incidence of the composite outcome was consistently higher in AI-ECG–graded high-risk cases across all European Society of Cardiology surgical risk and Revised Cardiac Risk Index strata. Conclusions: In this retrospective cohort, a multitask AI-ECG identified surgical candidates at low-risk for postoperative complications, for whom advanced cardiovascular testing demonstrated low diagnostic yield. Integrating AI-ECG with conventional risk tools may offer an exploratory strategy to optimize resource use and minimize redundant testing. Prospective studies are needed to confirm the clinical and economic benefits of AI-ECG as a screening tool.

  • Open Access

    Validation of AI-enhanced ECG image analysis for identifying extreme cardiac magnetic resonance metrics in a cross-ethnic UK biobank study

    by Yerin Kim, Haemin Lee, Hong-Mi Choi, In-Chang Hwang, Joonghee Kim, Ji Hyun Lee, Il-Young Oh, Heesun Lee, Su-Yeon Choi, Tae-Min Rhee, et al.

    Artificial intelligence enhanced electrocardiography (AI-ECG) has shown promise in detecting cardiac abnormalities, but validation against cardiac magnetic resonance (CMR), the reference standard for cardiac structure and function, remains limited. We evaluated the diagnostic performance of six AI-ECG models in identifying CMR-defined cardiac abnormalities in a population-based cohort. This cohort study included 38,804 UK Biobank participants with paired 12-lead ECG and CMR data. AI-ECG models, originally developed using Korean hospital-based datasets, were externally validated in this sample. Cardiac abnormalities were defined as the top 1% of CMR-derived values. Outcomes were the performance of AI-ECG in detecting left and right ventricular dysfunction, based on ejection fraction (QCG-LVD and QCG-RVD) and strain (ECG-LVGLS and ECG-RVGLS), and structural abnormalities including left ventricular hypertrophy (LVH) and left atrial enlargement (LAE). Model performance was assessed using the area under the receiver operating characteristic curve (AUC), with subgroup analyses by age, sex, and comorbidities. Among 38,804 participants (mean age 64.2 years; 48.2% male), AI-ECG models showed high accuracy for detecting left ventricular dysfunction (AUC 0.887 for QCG-LVD and 0.896 for ECG-LVGLS) and right ventricular dysfunction (AUC 0.778 for QCG-RVD and 0.825 for ECG-RVGLS). In a sub-cohort of 21,267 individuals, AUCs were 0.824 for detecting LVH and 0.883 for LAE. Subgroup analyses showed consistent performance, with higher accuracy among older individuals, males, and those with hypertension or ischemic heart disease. In this large multiethnic cohort, AI-ECG models demonstrated strong performance in detecting CMR-defined abnormalities, supporting their potential as a scalable, noninvasive screening tool for cardiovascular disease.


  • Open Access

    Artificial intelligence-enhanced ECG score for perioperative risk assessment in non-cardiac surgery

    by Hong-Mi Choi, Yerin Kim, Joonghee Kim, Jiesuck Park, Ji Hyun Lee, Yeonyee E Yoon, Il-Young Oh, In-Ae Song, Youngjin Cho

    Aims: The role of electrocardiography (ECG) has been limited in the preoperative risk evaluation in noncardiac surgery due to its low prognostic value. We aimed to evaluate the utility of an AI-enabled ECG (QCG-Critical score) in predicting 30-day postoperative mortality in non-cardiac surgery and compare its performance with traditional perioperative risk-assessment tools. Methods and results: A retrospective cohort of 46 135 adults who underwent non-cardiac surgery at a tertiary centre between 2020 and 2021 was analysed. Preoperative ECG images acquired within 30 days before surgery were used as input to previously developed CNN-based deep-learning algorithm to generate QCG-Critical score that reflects the risk for critical illness. The primary outcome was 30-day mortality, which occurred in 0.34% of patients. Individuals with QCG-Critical scores >40 had a markedly higher mortality rate of 11.7%. The QCG-Critical score demonstrated strong predictive performance for 30-day mortality (AUROC: 0.909), outperforming the ESC surgical category (0.728) and RCRI (0.725), and was comparable to the ASA classification (0.886). The performance of QCG-Critical score remained consistent across subgroups stratified by age, sex, emergency operation, anaesthesia type, and conventional risk groups. The QCG-Critical score also demonstrated good performance for predicting 7-day mortality (AUROC: 0.933), unplanned PCI (0.857), prolonged mechanical ventilation (0.829), and presumed heart failure (0.774). Conclusion: The preoperative QCG-Critical score accurately predicted postoperative mortality and other adverse outcomes, outperforming conventional risk-stratification tools. The QCG-Critical score may serve as a fast, accessible, and integrable tool for perioperative risk assessments in routine surgical care.


  • Open Access

    Screening for Left Ventricular Dysfunction in Sepsis Using a Smartphone ECG Analysis Application: A Multicenter Validation Study

    by Yun Seong Park, Joonghee Kim, You Hwan Jo, Woon Yong Kwon, Kyoung Jun Song, Hui Jai Lee, Youngjin Cho, Ji Eun Hwang

  • Open Access

    Artificial intelligence-driven ECG biomarkers for screening of large pericardial effusion

    by Moon-Seung Soh, Yangyoun Lee, GaYoung Kim, MokEun Park, Hong-Seok Lim, Joon-Han Shin, Youngjin Cho, Joonghee Kim

    Background: Pericardial effusion can progress to life-threatening cardiac tamponade when large or rapidly accumulating, yet early diagnosis is frequently delayed without sufficient clinical vigilance. Echocardiography is the gold standard but is not always accessible in emergency or resource-limited settings. Electrocardiographic (ECG) findings, including low QRS voltage and electrical alternans, may prompt clinical suspicion; however, their diagnostic value remains constrained by poor quantitative standardization and limited objectivity. Objective: This study aimed to evaluate the diagnostic performance of ECG Buddy™ , a smartphone-based artificial intelligence (AI) ECG analysis application, in detecting large pericardial effusion (LPE) and cardiac tamponade. Methods: We retrospectively reviewed adult patients (≥19 years) who underwent echocardiography between January 2013 and December 2023. LPE was defined as an effusion ≥2.0 cm in maximal diameter, and tamponade was defined as LPE with constrictive physiology or right atrium/right ventricle (RA/RV) diastolic collapse. An equal number of control patients without LPE were randomly sampled. Printed ECGs were analyzed using ECG Buddy™ , which generated a quantitative digital biomarker (QCG-LPE). Diagnostic performance was evaluated with receiver operating characteristic-area under the curve (ROC-AUC), sensitivity, specificity, and comparison with conventional biomarkers (Troponin I, NT-proBNP). Biomarker changes before and after pericardiocentesis (PCC) were also assessed. Results: A total of 770 patients (LPE=385, controls=385) were included, of whom 112 had tamponade and 158 underwent PCC. QCG-LPE levels differed markedly across groups: Normal 1.0 (IQR 0.3–3.5), LPE 37.8 (21.2–62.1), and Tamponade 61.8 (38.3–80.0) (p<0.001). QCG-LPE achieved an AUC of 0.980(95% CI, 0.972–0.988) for detecting LPE, with sensitivity 96.6% (95% CI, 94.3–98.7) and specificity 88.4% (95% CI, 84.9–91.6). For tamponade, the AUC was 0.880 (95% CI, 0.851–0.909), sensitivity 94.0% (95% CI, 88.1–98.8) and specificity 73.4% (95% CI, 69.7–77.0). QCG-LPE outperformed both Troponin I (AUC 0.731) and NT-proBNP (AUC 0.920) with a statistically significant difference (p<0.001). In patients undergoing PCC, biomarker values decreased significantly after the procedure (48.95 → 22.95, p<0.001), demonstrating responsiveness to therapeutic intervention. Conclusions: Digital ECG biomarker demonstrated excellent diagnostic accuracy for LPE and tamponade, validated against echocardiography as the gold standard. Its quantitative biomarker not only enables rapid screening but also reflects therapeutic response, suggesting potential for use as an accessible adjunct tool in emergency and resource-limited settings.

  • Open Access

    Multitask Artificial Intelligence–Based Electrocardiogram Tool for Preoperative Cardiac Testing in Noncardiac Surgery: Retrospective Cohort Study of Health Care Utilization and Costs

    by Hong-Mi Choi, Yerin Kim, Joonghee Kim, Jiesuck Park, In-Chang Hwang, Yun Young Choi, Ji Hyun Lee, Yeonyee E. Yoon, Il-Young Oh, Goo-Yeong Cho, et al.

    Background: Preoperative cardiovascular (CV) risk stratification is essential in non-cardiac surgery, but conventional testing is frequently overused, increasing costs without improving outcomes. Artificial intelligence (AI)-enabled electrocardiography (ECG) may enhance perioperative risk assessment by identifying patients at very low risk for adverse events. Objective: This study aimed to evaluate whether AI-ECG-based risk stratification could help maintain safety and decrease potentially avoidable preoperative CV testing, while reducing the associated costs, in patients undergoing non-cardiac surgery. Methods: We retrospectively analyzed 41,218 patients (46,135 ECG–surgery pairs) undergoing elective non-cardiac surgery at Seoul National University Bundang Hospital (2020–2021). An AI-ECG algorithm generated eight probability scores for cardiac conditions, classifying patients as low- or high-risk. Based on the performance and results of preoperative cardiovascular testing (transthoracic echocardiography, coronary computed tomography angiography, single-photon emission computed tomography, or coronary angiography), patients were classified as no test, negative test, or positive test. The primary endpoint was a 30-day composite of all-cause mortality, unplanned percutaneous coronary intervention, or prolonged mechanical ventilation (≥3 days). Results: AI-ECG classified 92.4% of patients as low-risk, with an event rate of 0.62% versus 6.04% in high-risk patients. Preoperative CV testing was performed in 11.8% of cases, with only 16.3% yielding positive findings. In AI-ECG low-risk patients, event rates were uniformly low (0.6–2.7%) regardless of testing, whereas in high-risk patients, rates were consistently high (5.3–6.2%), suggesting no additional prognostic value of conventional testing. An AI-ECG guided approach could reduce potentially avoidable preoperative testing by 35.7% while preserving safety. Integrating AI-ECG with established risk tools may better delineate patients who truly require preoperative CV testing while conserving medical resources. Conclusion: AI-enabled ECG reliably identified surgical candidates at low risk for postoperative complications, for whom additional CV testing may be potentially avoidable. Integrating AI-ECG with conventional risk tools may optimize resource use and minimize redundant testing without compromising outcomes. Prospective studies are needed to confirm clinical and economic benefits. Trial Registration: N/A

  • Open Access

    Multitask Artificial Intelligence–Based Electrocardiogram Tool for Preoperative Cardiac Testing in Noncardiac Surgery: Retrospective Cohort Study of Health Care Utilization and Costs (Preprint)

    by Hong-Mi Choi, Yerin Kim, Joonghee Kim, Jiesuck Park, In-Chang Hwang, Yun Young Choi, Ji Hyun Lee, Yeonyee E Yoon, Il-Young Oh, Goo-Yeong Cho, et al.

    BACKGROUND Preoperative cardiovascular risk stratification is essential in noncardiac surgery, but conventional testing is frequently overused, increasing costs without improving outcomes. Artificial intelligence (AI)–enabled electrocardiography (ECG) may enhance perioperative risk assessment by identifying surgical candidates at very low-risk for adverse events. OBJECTIVE This study aimed to evaluate whether AI-ECG–based risk stratification could help decrease low-yield preoperative cardiovascular testing and reduce associated costs, without an observed increase in postoperative adverse outcomes, in candidates for noncardiac surgery. METHODS We retrospectively analyzed 41,218 patients (46,135 ECG-surgery pairs) undergoing noncardiac surgery at Seoul National University Bundang Hospital (2020-2021). An AI-ECG algorithm generated eight probability scores for cardiac conditions, classifying cases as low- or high-risk. Based on the performance and results of preoperative cardiovascular testing (transthoracic echocardiography, coronary computed tomography angiography, single-photon emission computed tomography, or coronary angiography), cases were classified as no advanced cardiovascular imaging, negative-test, or positive-test. The primary end point was a 30-day composite of all-cause mortality and unplanned percutaneous coronary intervention. RESULTS AI-ECG classified 92.3% (42,599/46,135) of cases as low-risk, with a composite outcome rate of 0.2% (79/42,599) vs 2.9% (101/3536) in high-risk cases. Preoperative cardiovascular testing was performed in 11.8% (5458/46,135) of cases, with only 16.3% (892/5458) yielding positive findings. In AI-ECG low-risk cases, event rates were uniformly low (0.2%-0.4%) irrespective of whether advanced cardiovascular testing was performed, whereas in high-risk cases, rates were consistently high (2.6%-3.4%). The incidence of the composite outcome was consistently higher in AI-ECG–graded high-risk cases across all European Society of Cardiology surgical risk and Revised Cardiac Risk Index strata. CONCLUSIONS In this retrospective cohort, a multitask AI-ECG identified surgical candidates at low-risk for postoperative complications, for whom advanced cardiovascular testing demonstrated low diagnostic yield. Integrating AI-ECG with conventional risk tools may offer an exploratory strategy to optimize resource use and minimize redundant testing. Prospective studies are needed to confirm the clinical and economic benefits of AI-ECG as a screening tool. CLINICALTRIAL

  • Open Access

    Comparative Diagnostic Performance of a Multimodal Large Language Model Versus a Dedicated Electrocardiogram AI in Detecting Myocardial Infarction From Electrocardiogram Images: Comparative Study

    by Haemin Lee, Sooyoung Yoo, Joonghee Kim, Youngjin Cho, Dongbum Suh, Keehyuck Lee

    Background: Accurate and timely electrocardiogram (ECG) interpretation is critical for diagnosing myocardial infarction (MI) in emergency settings. Recent advances in multimodal large language models (LLMs), such as ChatGPT (OpenAI) and Gemini (Google DeepMind), have shown promise in clinical interpretation for medical imaging. However, whether these models analyze waveform patterns or simply rely on text cues remains unclear, underscoring the need for direct comparisons with dedicated ECG artificial intelligence (AI) tools. Objective: This study aimed to evaluate the diagnostic performance of ChatGPT and Gemini, a general-purpose LLM, in detecting MI from ECG images and to compare its performance with that of ECG Buddy (ARPI Inc) , a dedicated AI-driven ECG analysis tool. Methods: This retrospective study evaluated and compared AI models for classifying MI using a publicly available 12-lead ECG dataset from Pakistan, categorizing cases into MI-positive (239 images) and MI-negative (689 images). ChatGPT (GPT-4o, version November 20, 2024) and Gemini (Gemini 2.5 pro) were queried with 5 MI confidence options, whereas ECG Buddy for Microsoft Windows analyzed the images based on ST-elevation MI, acute coronary syndrome, and myocardial injury biomarkers. Results: Among 928 ECG recordings (239/928, 25.8% MI-positive), ChatGPT achieved an accuracy of 65.95% (95% CI 62.80‐69.00), area under the curve (AUC) of 57.34% (95% CI 53.44‐61.24), sensitivity of 36.40% (95% CI 30.30‐42.85), and specificity of 76.2% (95% CI 72.84‐79.33). With Gemini 2.5 Pro, accuracy dropped to 29.63% (95% CI 26.71‐32.69), AUC to 51.63% (95% CI 50.22‐53.04), and sensitivity rose to 97.07% (95% CI 94.06‐98.81), but specificity fell sharply to 6.24% (95% CI 4.55‐8.31). However, ECG Buddy reached an accuracy of 96.98% (95% CI 95.67‐97.99), AUC of 98.8% (95% CI 98.3‐99.43), sensitivity of 96.65% (95% CI 93.51‐98.54), and specificity of 97.10% (95% CI 95.55‐98.22). DeLong test confirmed that ECG Buddy significantly outperformed ChatGPT (all P <.001). In a qualitative error analysis of LLMs’ diagnostic explanations, GPT-4o produced fully accurate explanations in only 5% of cases (2/40), was partially accurate in 38% (15/40), and completely inaccurate in 58% (23/40). By contrast, Gemini 2.5 Pro yielded fully accurate explanations in 32% of cases (12/37), was partially accurate in 14% (5/37), and completely inaccurate in 54% (20/37). Conclusions: LLMs, such as ChatGPT and Gemini, underperform relative to specialized tools such as ECG Buddy in ECG image–based MI diagnosis. Further training may improve LLMs; however, domain-specific AI remains essential for clinical accuracy. The high performance of ECG Buddy underscores the importance of specialized models for achieving reliable and robust diagnostic outcomes.


  • Comparison of Extended Cryoballoon, Standard Cryoballoon, and Radiofrequency Catheter Ablation in Patients With Persistent AF

    by Min Soo Cho, Ji Hyun Lee, Il-Young Oh, Jung-Bok Lee, Minsoo Kim, Kee-Joon Choi, Youngjin Cho, Gi-Byoung Nam, Youngjin Cho, Jun Kim, et al.


  • Open Access

    Atrial fibrillation burden and symptom, quality of life, and healthcare resource utilization after cryoballoon ablation in persistent atrial fibrillation

    by Muhan Yeo, So-Ryoung Lee, JungMin Choi, Kyung-Yeon Lee, Hyo-Jeong Ahn, Soonil Kwon, Ji Hyun Lee, Youngjin Cho, Il-Young Oh, Hong-Euy Lim, et al.

    Aims: To investigate the relationship between continuously monitored atrial fibrillation (AF) burden after cryoballoon ablation (CBA) and improvements in AF-related symptoms, quality of life (QoL), and AF-related healthcare resource utilization (HCRU) in early persistent AF. Methods and results: This secondary analysis of the multicentre COOL-PER trial included patients with early persistent AF who underwent CBA and continuous monitoring via an implantable loop recorder. Post-CBA AF burden was defined as the percentage of time in AF between 9- and 12-month after CBA, categorized as <0.1%, 0.1 to <10%, and ≥10%. Symptom improvement was assessed using the European Heart Rhythm Association symptom score and QoL with the SF-36 survey. AF-related HCRU was defined as rhythm control interventions after the 90-day blanking period and cardiovascular-related hospitalizations or emergency room visits. Among 130 patients (mean baseline AF burden 77 ± 34%), AF burden significantly decreased post-CBA, with 50.0% achieving <0.1%, 28.5% in 0.1 to <10%, and 21.5% ≥ 10%. Symptom improvement rate was highest in the <0.1% group (89.2%), followed by the 0.1 to <10% (78.4%) and ≥10% (46.4%) groups (P < 0.001). Significant QoL improvement was observed in the <0.1% and 0.1 to <10% groups but not in the ≥10% group. Rhythm control interventions and cardiovascular-related hospitalizations or emergency room visits were more frequent in the ≥10% group (29%, 43%) than in the 0.1 to <10% (5.4%, 22%) and <0.1% (1.5%, 7.7%) groups (both P < 0.001). Conclusion: In early persistent AF, lower AF burden 1-year post-CBA was associated with greater AF-related symptom improvement, significant QoL enhancement, and reduced HCRU during follow-up.