Purpose: This study aimed to evaluate the role of hepatic resection in patients with objective responses after combined transarterial chemoembolization (TACE) and radiotherapy (RT) for hepatocellular carcinoma (HCC) with macroscopic vascular invasion (MVI). Materials and Methods: We retrospectively reviewed the patients treated with combined TACE and RT for HCC with MVI between 2010 and 2015. Some of the patients with objective responses underwent hepatic resection or liver transplantation; to investigate the impact of surgery, patients with objective responses who did not undergo surgery were selected as the control group. Survival outcomes were compared using a propensity score–based stabilized inverse probability of treatment weighting method. Results: Out of the 170 patients with objective responses after combined TACE and RT, 41 patients underwent surgery, including eight liver transplantations. The unweighted surgery group was younger and had a higher proportion of solitary tumors and unilateral vascular involvement. After adjustment, the 3-year overall survival (OS) rates were 61.0% and 28.6% in the surgery and non-surgery groups, respectively. The most important prognostic factor for OS was surgery (adjusted Cox hazard ratio [HR], 0.28; 95% confidence interval [CI], 0.17 to 0.46; p < 0.001). Complete response after TACE and RT (vs. partial response) was also a significant prognostic factor for OS (adjusted HR, 0.41; 95% CI, 0.27 to 0.61; p < 0.001). There was no surgical mortality. Four patients (9.8%) required additional surgery due to bleeding or graft failure. Conclusion Hepatic resection was significantly associated with improved OS in patients who showed objective responses after receiving combined TACE and RT for HCC with MVI.

Refractory iron deficiency anemia (IDA) in children should be investigated beyond dietary deficiency, particularly when accompanied by growth faltering or hypoalbuminemia. PTEN hamartoma tumor syndrome, which includes the Bannayan–Riley– Ruvalcaba syndrome (BRRS) spectrum, may cause diffuse gastrointestinal polyposis, resulting in chronic occult blood and protein loss. We report the case of a girl born in 2015 who presented at the age of two years with severe IDA (hemoglobin 5.0 g/dL, mean corpuscular volume 53.6 fL). Oral iron therapy was initiated, followed by intermittent intravenous iron infusions at progressively shorter intervals; however, the anemia recurred. Stool occult blood and fecal calprotectin test results were consistently negative during this period. Endoscopy and capsule endoscopy revealed diffuse gastrointestinal polyposis, including numerous small bowel polyps with ulcerative features. Genetic analysis revealed a heterozygous pathogenic variant in PTEN, NM_000314.8:c.389G＞A (p.Arg130Gln). Despite endoscopic polypectomy and oral nutritional supplementation, IDA, hypoalbuminemia (albumin, 2.5-3.2 g/dL), and growth faltering persisted. Extraintestinal findings included macrocephaly, lipomatous masses, a subcutaneous hamartoma, and brain magnetic resonance imaging findings consistent with BRRS. Sirolimus therapy was initiated in September 2025, targeting a trough level of 6-10 ng/mL. By 6 months, hemoglobin stabilized at 12.9 g/dL, albumin normalized to 4.4 g/dL, weight increased by 5.1 kg, body mass index improved from below the 3rd percentile to the 10th percentile, and intravenous iron was no longer required. This case suggests that sirolimus may be beneficial in the treatment of diffuse PTEN-related gastrointestinal polyposis when endoscopic management is insufficient.

Purpose: This study aimed to develop a magnetic resonance imaging (MRI)–based radiomics model to predict high-risk pathologic features for lung adenocarcinoma: micropapillary and solid pattern (MPsol), spread through air space, and poorly differentiated patterns. Materials and Methods: As a prospective study, we screened clinical N0 lung cancer patients who were surgical candidates and had undergone both 18F-fluorodeoxyglucose (FDG) positron emission tomography–computed tomography (PET/CT) and chest CT from August 2018 to January 2020. We recruited patients meeting our proposed imaging criteria indicating high-risk, that is, poorer prognosis of lung adenocarcinoma, using CT and FDG PET/CT. If possible, these patients underwent an MRI examination from which we extracted 77 radiomics features from T1-contrast-enhanced and T2-weighted images. Additionally, patient demographics, maximum standardized uptake value on FDG PET/CT, and the mean apparent diffusion coefficient value on diffusion-weighted image, were considered together to build prediction models for high-risk pathologic features. Results: Among 616 patients, 72 patients met the imaging criteria for high-risk lung cancer and underwent lung MRI. The magnetic resonance (MR)–eligible group showed a higher prevalence of nodal upstaging (29.2% vs. 4.2%, p < 0.001), vascular invasion (6.5% vs. 2.1%, p=0.011), high-grade pathologic features (p < 0.001), worse 4-year disease-free survival (p < 0.001) compared with non-MR-eligible group. The prediction power for MR-based radiomics model predicting high-risk pathologic features was good, with mean area under the receiver operating curve (AUC) value measuring 0.751-0.886 in test sets. Adding clinical variables increased the predictive performance for MPsol and the poorly differentiated pattern using the 2021 grading system (AUC, 0.860 and 0.907, respectively). Conclusion Our imaging criteria can effectively screen high-risk lung cancer patients and predict high-risk pathologic features by our MR-based prediction model using radiomics.

Objectives: Developing large language models (LLMs) in biomedicine requires access to high-quality training and alignment tuning datasets. However, publicly available Korean medical preference datasets are scarce, hindering the advancement of Korean medical LLMs. This study constructs and evaluates the efficacy of the Korean Medical Preference Dataset (KoMeP), an alignment tuning dataset constructed with an automated pipeline, minimizing the high costs of human annotation. Methods: KoMeP was generated using the DAHL score, an automated hallucination evaluation metric. Five LLMs (Dolly-v2-3B, MPT-7B, GPT-4o, Qwen-2-7B, Llama-3-8B) produced responses to 8,573 biomedical examination questions, from which 5,551 preference pairs were extracted. Each pair consisted of a “chosen” response and a “rejected” response, as determined by their DAHL scores. The dataset was evaluated when trained through two different alignment tuning methods, direct preference optimization (DPO) and odds ratio preference optimization (ORPO) respectively across five different models. The KorMedMCQA benchmark was employed to assess the effectiveness of alignment tuning. Results: Models trained with DPO consistently improved KorMedMCQA performance; notably, Llama-3.1-8B showed a 43.96% increase. In contrast, ORPO training produced inconsistent results. Additionally, English-to-Korean transfer learning proved effective, particularly for English-centric models like Gemma-2, whereas Korean-to-English transfer learning achieved limited success. Instruction tuning with KoMeP yielded mixed outcomes, which suggests challenges in dataset formatting. Conclusions KoMeP is the first publicly available Korean medical preference dataset and significantly improves alignment tuning performance in LLMs. The DPO method outperforms ORPO in alignment tuning. Future work should focus on expanding KoMeP, developing a Korean-native dataset, and refining alignment tuning methods to produce safer and more reliable Korean medical LLMs.

Objectives: Developing large language models (LLMs) in biomedicine requires access to high-quality training and alignment tuning datasets. However, publicly available Korean medical preference datasets are scarce, hindering the advancement of Korean medical LLMs. This study constructs and evaluates the efficacy of the Korean Medical Preference Dataset (KoMeP), an alignment tuning dataset constructed with an automated pipeline, minimizing the high costs of human annotation. Methods: KoMeP was generated using the DAHL score, an automated hallucination evaluation metric. Five LLMs (Dolly-v2-3B, MPT-7B, GPT-4o, Qwen-2-7B, Llama-3-8B) produced responses to 8,573 biomedical examination questions, from which 5,551 preference pairs were extracted. Each pair consisted of a “chosen” response and a “rejected” response, as determined by their DAHL scores. The dataset was evaluated when trained through two different alignment tuning methods, direct preference optimization (DPO) and odds ratio preference optimization (ORPO) respectively across five different models. The KorMedMCQA benchmark was employed to assess the effectiveness of alignment tuning. Results: Models trained with DPO consistently improved KorMedMCQA performance; notably, Llama-3.1-8B showed a 43.96% increase. In contrast, ORPO training produced inconsistent results. Additionally, English-to-Korean transfer learning proved effective, particularly for English-centric models like Gemma-2, whereas Korean-to-English transfer learning achieved limited success. Instruction tuning with KoMeP yielded mixed outcomes, which suggests challenges in dataset formatting. Conclusions KoMeP is the first publicly available Korean medical preference dataset and significantly improves alignment tuning performance in LLMs. The DPO method outperforms ORPO in alignment tuning. Future work should focus on expanding KoMeP, developing a Korean-native dataset, and refining alignment tuning methods to produce safer and more reliable Korean medical LLMs.

Purpose: This study aimed to develop a magnetic resonance imaging (MRI)–based radiomics model to predict high-risk pathologic features for lung adenocarcinoma: micropapillary and solid pattern (MPsol), spread through air space, and poorly differentiated patterns. Materials and Methods: As a prospective study, we screened clinical N0 lung cancer patients who were surgical candidates and had undergone both 18F-fluorodeoxyglucose (FDG) positron emission tomography–computed tomography (PET/CT) and chest CT from August 2018 to January 2020. We recruited patients meeting our proposed imaging criteria indicating high-risk, that is, poorer prognosis of lung adenocarcinoma, using CT and FDG PET/CT. If possible, these patients underwent an MRI examination from which we extracted 77 radiomics features from T1-contrast-enhanced and T2-weighted images. Additionally, patient demographics, maximum standardized uptake value on FDG PET/CT, and the mean apparent diffusion coefficient value on diffusion-weighted image, were considered together to build prediction models for high-risk pathologic features. Results: Among 616 patients, 72 patients met the imaging criteria for high-risk lung cancer and underwent lung MRI. The magnetic resonance (MR)–eligible group showed a higher prevalence of nodal upstaging (29.2% vs. 4.2%, p < 0.001), vascular invasion (6.5% vs. 2.1%, p=0.011), high-grade pathologic features (p < 0.001), worse 4-year disease-free survival (p < 0.001) compared with non-MR-eligible group. The prediction power for MR-based radiomics model predicting high-risk pathologic features was good, with mean area under the receiver operating curve (AUC) value measuring 0.751-0.886 in test sets. Adding clinical variables increased the predictive performance for MPsol and the poorly differentiated pattern using the 2021 grading system (AUC, 0.860 and 0.907, respectively). Conclusion Our imaging criteria can effectively screen high-risk lung cancer patients and predict high-risk pathologic features by our MR-based prediction model using radiomics.

Purpose: This study aimed to develop a magnetic resonance imaging (MRI)–based radiomics model to predict high-risk pathologic features for lung adenocarcinoma: micropapillary and solid pattern (MPsol), spread through air space, and poorly differentiated patterns. Materials and Methods: As a prospective study, we screened clinical N0 lung cancer patients who were surgical candidates and had undergone both 18F-fluorodeoxyglucose (FDG) positron emission tomography–computed tomography (PET/CT) and chest CT from August 2018 to January 2020. We recruited patients meeting our proposed imaging criteria indicating high-risk, that is, poorer prognosis of lung adenocarcinoma, using CT and FDG PET/CT. If possible, these patients underwent an MRI examination from which we extracted 77 radiomics features from T1-contrast-enhanced and T2-weighted images. Additionally, patient demographics, maximum standardized uptake value on FDG PET/CT, and the mean apparent diffusion coefficient value on diffusion-weighted image, were considered together to build prediction models for high-risk pathologic features. Results: Among 616 patients, 72 patients met the imaging criteria for high-risk lung cancer and underwent lung MRI. The magnetic resonance (MR)–eligible group showed a higher prevalence of nodal upstaging (29.2% vs. 4.2%, p < 0.001), vascular invasion (6.5% vs. 2.1%, p=0.011), high-grade pathologic features (p < 0.001), worse 4-year disease-free survival (p < 0.001) compared with non-MR-eligible group. The prediction power for MR-based radiomics model predicting high-risk pathologic features was good, with mean area under the receiver operating curve (AUC) value measuring 0.751-0.886 in test sets. Adding clinical variables increased the predictive performance for MPsol and the poorly differentiated pattern using the 2021 grading system (AUC, 0.860 and 0.907, respectively). Conclusion Our imaging criteria can effectively screen high-risk lung cancer patients and predict high-risk pathologic features by our MR-based prediction model using radiomics.

Objectives: Developing large language models (LLMs) in biomedicine requires access to high-quality training and alignment tuning datasets. However, publicly available Korean medical preference datasets are scarce, hindering the advancement of Korean medical LLMs. This study constructs and evaluates the efficacy of the Korean Medical Preference Dataset (KoMeP), an alignment tuning dataset constructed with an automated pipeline, minimizing the high costs of human annotation. Methods: KoMeP was generated using the DAHL score, an automated hallucination evaluation metric. Five LLMs (Dolly-v2-3B, MPT-7B, GPT-4o, Qwen-2-7B, Llama-3-8B) produced responses to 8,573 biomedical examination questions, from which 5,551 preference pairs were extracted. Each pair consisted of a “chosen” response and a “rejected” response, as determined by their DAHL scores. The dataset was evaluated when trained through two different alignment tuning methods, direct preference optimization (DPO) and odds ratio preference optimization (ORPO) respectively across five different models. The KorMedMCQA benchmark was employed to assess the effectiveness of alignment tuning. Results: Models trained with DPO consistently improved KorMedMCQA performance; notably, Llama-3.1-8B showed a 43.96% increase. In contrast, ORPO training produced inconsistent results. Additionally, English-to-Korean transfer learning proved effective, particularly for English-centric models like Gemma-2, whereas Korean-to-English transfer learning achieved limited success. Instruction tuning with KoMeP yielded mixed outcomes, which suggests challenges in dataset formatting. Conclusions KoMeP is the first publicly available Korean medical preference dataset and significantly improves alignment tuning performance in LLMs. The DPO method outperforms ORPO in alignment tuning. Future work should focus on expanding KoMeP, developing a Korean-native dataset, and refining alignment tuning methods to produce safer and more reliable Korean medical LLMs.

Background/Aims: Metabolic dysfunction-associated steatotic liver disease (MASLD) is characterized by fat accumulation in the liver. MASLD encompasses both steatosis and MASH. Since MASH can lead to cirrhosis and liver cancer, steatosis and MASH must be distinguished during patient treatment. Here, we investigate the genomes, epigenomes, and transcriptomes of MASLD patients to identify signature gene set for more accurate tracking of MASLD progression. Methods: Biopsy-tissue and blood samples from patients with 134 MASLD, comprising 60 steatosis and 74 MASH patients were performed omics analysis. SVM learning algorithm were used to calculate most predictive features. Linear regression was applied to find signature gene set that distinguish the stage of MASLD and to validate their application into independent cohort of MASLD. Results: After performing WGS, WES, WGBS, and total RNA-seq on 134 biopsy samples from confirmed MASLD patients, we provided 1,955 MASLD-associated features, out of 3,176 somatic variant callings, 58 DMRs, and 1,393 DEGs that track MASLD progression. Then, we used a SVM learning algorithm to analyze the data and select the most predictive features. Using linear regression, we identified a signature gene set capable of differentiating the various stages of MASLD and verified it in different independent cohorts of MASLD and a liver cancer cohort. Conclusions We identified a signature gene set (i.e., CAPG, HYAL3, WIPI1, TREM2, SPP1, and RNASE6) with strong potential as a panel of diagnostic genes of MASLD-associated disease.

Background: The Korea Expert Committee on Immunization Practices (KECIP) is a key advisory body the government to develop guidelines and provide technical advisory activities on immunization policies in Korea. A recent policy study, inspired by global best practices, aims to enhance KECIP's functionality for providing timely and transparent recommendations in the face of evolving vaccine science and emerging infectious diseases like COVID-19. Methods: This study reviewed the current status of KECIP and collected expert opinions through surveys and consultations. Among the 40 panel members who were surveyed, 19 responded to a questionnaire specifically designed to assess the potential areas of improvement within KECIP. Results: The majority of respondents favored maintaining the current member count and emphasized the need for a subcommittee. Opinions varied on issues such as the length of KECIP’s term, the representation of vaccine manufacturers’ perspectives, and the chairperson’s role. However, there was a consensus on the importance of expertise, transparency, and fair proceedings within the committee. Conclusion This study underscores the pivotal role of KECIP in shaping national immunization policies, emphasizing the necessity for informed guidance amidst evolving vaccine science and emerging infectious diseases. Furthermore, it stressed the importance of enhancing KECIP’s capacity to effectively address evolving public health challenges and maintain successful immunization programs in South Korea.