1.Evaluation of the Performance of Advanced Large Language Models in Laboratory Medicine Using Residency Examinations
Kiwook JUNG ; Hyun Jin KIM ; Sunghwan SHIN ; Wookeun LEE ; Jun Hyung LEE ; Hee Sue PARK ; Qute CHOI
Annals of Laboratory Medicine 2026;46(3):327-337
Background:
Recent advancements in large language models (LLMs) have accelerated their integration into clinical domains, including laboratory medicine. The performance of LLMs in answering board-level laboratory medicine questions has not been comprehensively evaluated. Given the importance of diagnostic accuracy in this field, rigorous and objective evaluations of LLM capabilities are essential.
Methods:
We assessed 12 LLMs from OpenAI, Anthropic, and Google using 320 Korean Residency Examination questions (2021–2024) spanning six laboratory medicine subspecialties. Standardized prompts were provided via their application programming interfaces under deterministic settings (temperature = 0). Questions were administered thrice to assess response reproducibility. Outputs were compared with validated answers and analyzed for accuracy, reasoning quality, and error typology.
Results:
Google’s Gemini 2.0 Pro achieved the highest accuracy (80.0%), followed by OpenAI’s GPT-4.5 (77.2%) and Anthropic’s Claude 3.7 Sonnet (74.1%). Accuracy decreased as the difficulty of questions increased (78.0% for easy vs. 45.1% for challenging). Subspecialty performance varied. Al models underperformed on questions on transfusion medicine (mean accuracy: 38.8%), primarily because of limitations in domain-specific and regional knowledge representations. Incorrect answers primarily resulted from reasoning errors. Reproducibility exceeded 95% for most models; however, some residual non-determinism appeared even with greedy decoding (temperature = 0).
Conclusions
LLMs demonstrated substantial potential for integration into laboratory medicine, particularly in clinical chemistry and immunology. Performance inconsistencies (particularly for high-difficulty questions) and knowledge gaps (notably for transfusion medicine) highlight the necessity for further development—potentially including domain-specific fine-tuning and retrieval-augmented generation integration—and robust expert oversight before clinical application.
2.Three-year outcomes of a prospective, multicenter study of rotational atherectomy with antirestenotic therapy for infrainguinal arterial disease
Sungsin CHO ; Hyung-Kee KIM ; Woo-Sung YUN ; Ui Jun PARK ; Sang Su LEE ; Jaehoon LEE ; Hong-Pil HWANG ; Jin Hyun JOH
Annals of Surgical Treatment and Research 2026;110(3):180-187
Purpose:
Atherosclerotic plaques in peripheral arterial disease (PAD) include fatty, mixed, and calcified types. Plaque burden is significantly associated with restenosis, reintervention, and amputation-free survival. Rotational and aspirational atherectomy (RAA) may effectively remove such plaques. This study aimed to evaluate long-term outcomes of RAA for infrainguinal PAD.
Methods:
Patients with infrainguinal lesions underwent revascularization using the Jetstream Atherectomy System (Boston Scientific). This 60-month extension assessed primary patency rate (PPR) and clinically driven target lesion revascularization (CD-TLR). Kaplan-Meier curves were used for survival analysis; P < 0.05 was considered statistically significant.
Results:
A total of 150 patients (mean age, 70.9 years; male, 86.0%; 65.4% with diabetes) were enrolled. The mean lesion length was 15.8 cm, with 74.0% occlusions and 47.3% severe calcification. Lesions were sclerotic (72.4%), thrombosclerotic (13.4%), thrombotic (9.4%), or in-stent (4.7%). A drug-coated balloon (DCB) was used in 85.5% of cases. PPR at 1, 3, and 5 years was 84.1%, 68.1%, and 58.5%, respectively. CD-TLR rates were 93.0%, 81.5%, and 67.4%, respectively. The benefit of DCB was sustained through 3 years but attenuated thereafter, highlighting the need for extended follow-up in infrainguinal interventions.
Conclusion
RAA demonstrated durable 5-year patency and safety outcomes. Device type, DCB use, lesion morphology, and calcium grade did not significantly influence long-term results. Lesion complexity remains the primary predictor of clinical outcome. Despite the complexity of infrainguinal lesions, the use of RAA demonstrated sustained patency through 3 years, with lesion complexity (particularly TASC classification) emerging as the most critical predictor of long-term success.
3.Radiation-Associated Heart Disease in Korean Women after Radiotherapy for Breast Cancer: Insights from the National Health Insurance Service Database
Jun-Hyuk LEE ; Jimin PARK ; Tae Hyung KIM
Cancer Research and Treatment 2026;58(2):534-543
Purpose:
This study investigated the risk of radiation-associated heart disease (RAHD) in Korean women treated with radiotherapy (RT) for breast cancer (BC) using data from the National Health Insurance Service database.
Materials and Methods:
A retrospective cohort analysis was conducted on 65,188 patients with BC treated with RT between 2009 and 2014 and 325,940 controls without BC or prior coronary artery disease (CAD), with 1:5 exact matching by age, type 2 diabetes mellitus, hypertension, and dyslipidemia status. CAD encompassed both incident events and fatal events. Competing risk analysis was conducted to estimate subdistribution hazard ratio (HR) with 95% confidence interval (CI) for CAD, setting mortality from non-CAD causes as a competing risk.
Results:
During the mean 9.9 years of follow-up period, 3,852 (1.0%) CAD and 20,999 (5.4%) death from non-CAD causes were reported. Compared to controls, participants with BC who received RT had a significantly lower risk of CAD incidence. HR (95% CI) for CAD in the BC with RT group was 0.66 (0.60-0.73, p < 0.001). On the other hand, HR (95% CI) for mortality from non-CAD causes was 3.57 (3.48-3.67, p < 0.001).
Conclusion
In this large population-based cohort study, breast cancer patients who received RT did not show an increased incidence of CAD compared with the general population without breast cancer. Individual-level dosimetric data and longer follow-up are needed to clarify the independent risk.
4.Comparative Study of Mastectomy Using Conventional Techniques, Multiport and Single-Port Robotic Surgical Systems
Jeea LEE ; Jieon GO ; Suk Jun LEE ; Yonghan KWON ; Nam Hee KIM ; Jee Ye KIM ; Hyung Seok PARK
Cancer Research and Treatment 2026;58(2):492-500
Purpose:
Oncologic and surgical outcomes of robot-assisted nipple-sparing mastectomy (RNSM) compared to conventional nipple-sparing mastectomy (CNSM) is under investigation. This study compared the clinical outcomes of recurrence-free survival and postoperative complication after RNSM and CNSM.
Materials and Methods:
We retrospectively reviewed data of 401 patients who underwent da Vinci Si/Xi/SP-assisted RNSM or CNSM with immediate reconstruction between November 2016 and November 2020 at a single institute. Oncological outcomes were collected until March 2022. Primary endpoints were long-term outcomes, such as local recurrence, distant metastasis, disease-free survival, overall survival, and postoperative complications, while secondary endpoints were pathology results, and oncological outcomes.
Results:
Patients underwent RNSM (n=162) or CNSM (n=239). Of RNSM cases, 9 (5.6%) were performed using the da Vinci Si System, 96 (59.3%) using the da Vinci Xi System, and 57 (35.2%) using the da Vinci SP System. No significant difference in recurrence-free survival was found between the RNSM and CNSM group, and both groups had a median follow-up of 37 months. The recurrence rate in RNSM patients after a median follow-up of 24.5 months was 3.8%, compared with 5.9% in CNSM patients after a median follow-up of 42 months. No difference in recurrence was seen among RNSM patients with respect to surgical systems (multiport vs. SP, p =0.136). In addition, grade III postoperative complication rate was lower in patients with RNSM than in those with CNSM. Transfusion was only applied in 6.2% of patients.
Conclusion
Robot-assisted surgical systems can be safely used to perform nipple-sparing mastectomy in patients with early breast cancer.
5.Association of Physical Activity with Dementia Risk in Cancer Survivors: A Korean Nationwide Cohort Study
Su Kyoung LEE ; Minji HAN ; Sangwoo PARK ; Sun Jae PARK ; Jihun SONG ; Hye Jun KIM ; Jaewon KIM ; Hyeokjong LEE ; Hyun-Young SHIN ; Kyae Hyung KIM ; Sang Min PARK
Cancer Research and Treatment 2026;58(1):48-60
Purpose:
This study aimed to investigate the impact of physical activity on dementia risk among cancer survivors in South Korea.
Materials and Methods:
This retrospective, population-based cohort study included 344,152 cancer survivors identified from the National Health Insurance Service database in South Korea. The mean follow-up time was 5.81 years. Different levels of physical activity post-cancer diagnosis, ranging from inactive to highly active, were assessed. The primary outcome was the incidence of overall dementia, Alzheimer’s disease, and vascular dementia. Secondary outcomes included dementia risk stratified by cancer type and treatment (chemotherapy and radiation).
Results:
Of the total participants, 24,363 (7.08%) developed dementia. The risk of overall dementia decreased sequentially across the exercise groups compared to the inactive group: insufficiently active (adjusted hazard ratio [HR], 0.89; 95% confidence interval [CI], 0.86 to 0.92), active (adjusted HR, 0.85; 95% CI, 0.83 to 0.88), and highly active (adjusted HR, 0.79; 95% CI, 0.76 to 0.82). This inverse relationship between exercise and dementia risk was statistically significant across various cancer types and was consistent regardless of age, comorbidities, and whether or not excluding the first 1, 2 years.
Conclusion
Among cancer survivors in South Korea, increased physical activity post-diagnosis was associated with a significantly lower risk of dementia. These findings underscore the importance of promoting physical activity in cancer survivors for cognitive health.
6.Detection Ability of Quality of Life Changes and Responsiveness of the KOQUSS-40 and the EORTC QLQ-C30/STO22 in Patients Who Underwent Gastrectomy: A Prospective Comparative Study
Bang Wool EOM ; Keun Won RYU ; Ji Yeong AN ; Yun-Suhk SUH ; In CHO ; Sung Geun KIM ; Ji-Ho PARK ; Hoon HUR ; Hyung-Ho KIM ; Sang-Hoon AHN ; Sun-Hwi HWANG ; Hong Man YOON ; Ki Bum PARK ; Hyoung-Il KIM ; In-Gyu KWON ; Han-Kwang YANG ; Byoung-Jo SUH ; Sang-Ho JEONG ; Tae-Han KIM ; Oh Kyoung KWON ; Hye-Seong AHN ; Ji Yeon PARK ; Ki Young YOON ; Myoung Won SON ; Seong-Ho KONG ; Young-Gil SON ; Geum Jong SONG ; Jong Hyuk YUN ; Jung-Min BAE ; Do Joong PARK ; Sol LEE ; Jun-Young YANG ; Kyung Won SEO ; You-Jin JANG ; So Hyun KANG ; Joongyub LEE ; Hyuk-Joon LEE ;
Cancer Research and Treatment 2026;58(1):221-231
Purpose:
The aim of this study is to compare the detection ability of quality of life (QoL) changes and responsiveness of the KOrean QUality of life in Stomach cancer patients Study group (KOQUSS)-40 and European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ).
Materials and Methods:
A multicenter prospective observational study was conducted to evaluate QoL changes after various gastrectomies between January 2021 and April 2022. Participants were instructed to complete the KOQUSS-40 and EORTC QLQ-C30/STO22 preoperatively and at 1, 3, 6, and 12 months postoperatively. QoL changes over time and QoL responsiveness were assessed for each questionnaire.
Results:
Data from 491 patients who underwent curative gastrectomy for gastric cancer at 22 institutions were analyzed. The summary scores of the KOQUSS-40 and EORTC QLQ-STO22 showed significant differences between the total and proximal gastrectomy groups (p=0.044 and p=0.038, respectively), but no difference was observed for the EORTC QLQ-C30. Dysphagia on the KOQUSS-40 was significantly different between the total and proximal gastrectomy groups (p=0.031); however, dysphagia on the EORTC QLQ-STO22 did not differ. The responsiveness of the KOQUSS-40 was similar to that of the EORTC QLQ in patients who experienced ≥ 10% body weight loss, but approximately 10% less in patients receiving adjuvant chemotherapy than the EORTC QLQ.
Conclusion
KOQUSS-40 has several advantages over EORTC QLQ-C30/STO22 when comparing QoL between the total and proximal gastrectomy groups. The findings provide information for researchers investigating the QoL of patients who have undergone curative gastrectomy for gastric cancer.
7.Survival Rates of Patients with Gastric Cancer According to Age and Sex: A Large-Scale Study Using Data from 14,739 Patients
Yonghoon CHOI ; Nayoung KIM ; Ji Hyun KIM ; Hyeong Ho JO ; Hyeon Jeong OH ; Hye Seung LEE ; Yu Kyung JUN ; Hyuk YOON ; Cheol Min SHIN ; Young Soo PARK ; Dong Ho LEE ; So Hyun KANG ; Young Suk PARK ; Sang-Hoon AHN ; Yun-Suhk SUH ; Do Joong PARK ; Hyung Ho KIM ; Ji-Won KIM ; Jin Won KIM ; Keun-Wook LEE ; Won CHANG ; Yoon Jin LEE ; Kyoung Ho LEE ; Young Hoon KIM
Cancer Research and Treatment 2026;58(1):252-263
Purpose:
The male predominance in the incidence of gastric cancer (GC) is established; however, sex differences in the prognosis of GC remain controversial. As such, this study analyzed the prognosis of patients with GC based on age and sex.
Materials and Methods:
Data from 14,739 patients diagnosed with GC at Seoul National University Bundang Hospital between 2003 and 2023 were analyzed. Baseline characteristics, histological types of GC, overall and GC-specific survival rates (age and stage stratification), and associated risk factors were analyzed.
Results:
Females were significantly younger (p < 0.001) and exhibited more gastric body cancers (p < 0.001) and tumors with diffuse-type or poorly differentiated histology (p < 0.001) than males. Females exhibited an advantage over males in terms of overall survival (p=0.004), but not in GC-specific survival. However, age stratification revealed significant sex differences, that females < 50 years of age exhibited survival disadvantages (p < 0.001); however, this trend was reversed with age, and females > 60 years exhibited survival advantages (p < 0.001) for both overall and GC-specific survival. This may be explained by the lower ratio of diffuse-type GC as females age. Furthermore, in the analysis according to stage, females with stage IV disease exhibited significant survival disadvantages, with significantly younger age and a higher proportion of diffuse-type GC which exhibits aggressive features, resulting in poorer survival than in males.
Conclusion
Age and stage stratification revealed significant differences in survival between the sexes, which can be helpful for public health strategies.
8.Comprehensive Characterization of Spastic Paraplegia in Korean Patients: A Single-Center Experience over Two Decades
Yunjung CHOI ; Soo-Hyun KIM ; Sung Jun AHN ; Eun Kyoung OH ; Jeong Hee CHO ; Ha Young SHIN ; Seung Woo KIM ; Young-Chul CHOI ; Hyung Jun PARK
Yonsei Medical Journal 2026;67(1):34-41
Purpose:
Hereditary spastic paraplegia (HSP) refers to a group of genetic neurodegenerative diseases marked by gradually worsening spasticity and hyperreflexia in the lower extremities. This study aimed to describe the clinical and genetic characteristics of Korean patients with spastic paraplegia.
Materials and Methods:
We retrospectively reviewed medical records of 69 patients with spastic paraplegia from 54 unrelated families between 2002 and 2024. Genetic, clinical, electrophysiological, and radiological features were comprehensively analyzed.
Results:
Causative genes were identified in 34 (63%) of 54 unrelated families; SPAST, detected in 26 families, was the most prevalent. Seven novel pathogenic variants were identified. Clinically, the median age of symptom onset was 25 years [14.0–37.0]. Out of 69 patients with spastic paraplegia, 51 (74%) presented with the pure form of spastic paraplegia, which included all patients with SPG4. Spastic gait was a universal feature in all patients. Urinary dysfunction was present in 42 (61%) patients. Additional neurologic manifestations included peripheral neuropathy 9 (13%), cognitive impairment 5 (7%), upper limb weakness 4 (6%), dysarthria 4 (6%), dysphagia 3 (4%), ataxia 3 (4%), and scoliosis 1 (3%). Brain MRI findings demonstrated a thin corpus callosum in two patients with SPG11; all patients with SPG4 had normal findings. Spine MRI revealed spinal cord atrophy in 16 (27%) patients, including 6 (21%) patients with SPG4.
Conclusion
The study comprehensively reviewed genetic and clinical spectra of spastic paraplegia in Korean patients, emphasizing the predominance of SPAST as the causative gene and underscoring the genetic and phenotypic heterogeneity of spastic paraplegia.
10.Estimating the Prevalence of Autosomal Recessive Neuromuscular Diseases in the Korean Population
Soo-Hyun KIM ; Yunjung CHOI ; Young-Chul CHOI ; Seung Woo KIM ; Ha Young SHIN ; Hyung Jun PARK
Journal of Korean Medical Science 2025;40(19):e68-
Background:
Genetic neuromuscular diseases (NMDs) are a heterogeneous group of conditions that primarily affect the peripheral nerves, muscles, and neuromuscular junctions. This study was performed to identify pathogenic or likely pathogenic variants (PLPVs), calculate carrier frequencies, and predict the genetic prevalence of autosomal recessive-NMDs (AR-NMDs) in a Korean population.
Methods:
In total, 267 genes were associated with AR-NMDs. We analyzed genetic variants from 984 Korean whole genomes and identified PLPVs to assess the carrier frequency and genetic prevalence of the variants.
Results:
We identified 165 PLPVs, including 75 literature verified and 90 manually verified variants. Most PLPVs in AR-NMD genes were frameshifts (61, 37.0%), followed by nonsense (36, 21.8%), missense (35, 21.2%), and splice variants (28, 17.0%). The carrier frequency of the AR-NMDs was 27.1%. DYSF exhibited the highest carrier frequency (1.63%), followed by GAA (1.55%), HEXB (1.53%), PREPL (0.76%), NEB (0.66%), ADSS1 (0.65%), ALPK3 (0.65%), and CHRNG (0.65%). The predicted genetic prevalence of AR-NMDs in the Korean population was 38.0 cases per 100,000 individuals. DYSF (6.7 cases per 100,000 individuals) showed the highest genetic prevalence. The variant with the highest allele frequency was c.1250C>T in HEXB at 0.00764, followed by c.[752T>C; c.761C>T] in GAA at 0.00505, and c.2055+2T>G in DYSF at 0.00437.
Conclusion
Our study suggests that 27.1% of the Korean population are healthy carriers of at least one AR-NMD causing PLPV, revealing the genetic prevalence of NMDs in the Korean population.

Result Analysis
Print
Save
E-mail