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.

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: Teicoplanin is a glycopeptide antimicrobial that treats serious invasive infections caused by gram-positive bacteria, such as the methicillin-resistant Staphylococcus aureus. Despite some comparable advantages, there is no guideline or clinical recommendation for teicoplanin in the pediatric population, unlike vancomycin where abundant studies and the recently revised guideline on therapeutic drug level monitoring (TDM) exist. Methods: The systematic review was performed in accordance with the preferred reporting items for systematic reviews. Two authors (JSC and SHY) searched PubMed, Embase, and Cochrane Library databases using relevant terms independently. Results: Fourteen studies were finally included with a total of 1,380 patients. TDM was available in 2,739 samples collected in the nine studies. Dosing regimens varied widely, and eight studies used recommended dosing regimens. Timing for measuring TDM was mostly 72–96 hours or longer after the initiation of the first dose, which was expected to be a steadystate. The majority of studies had target trough levels of 10 µg/mL or above. Three studies reported that the clinical efficacy and treatment success rate of teicoplanin was 71.4%, 87.5%, and 88%. Adverse events associated with teicoplanin use were described in six studies with a focus on renal and/or hepatic impairment. Except for one study, no significant relation was noted between the incidence of adverse events and trough concentration. Conclusion Current evidence on teicoplanin trough levels in pediatric populations is insufficient due to heterogeneity. However, target trough levels with favorable clinical efficacy are achievable by recommended dosing regimen in the majority of patients.

Neutralizing antibodies targeted at the receptor-binding domain of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike protein have been developed and now under evaluation in clinical trials. The US Food and Drug Administration currently issued emergency use authorizations for neutralizing monoclonal antibodies in non-hospitalized patients with mild to moderate coronavirus disease 2019 (COVID-19) who are at high risk for progressing to severe disease and/or hospitalization. In terms of this situation, there is an urgent need to investigate the clinical aspects and to develop strategies to deploy them effectively in clinical practice. Here we provide guidance for the use of anti-SARS-CoV-2 monoclonal antibodies for the treatment of COVID-19 based on the latest evidence.

Despite the global effort to mitigate the spread, coronavirus disease 2019 (COVID-19) has become a pandemic that took more than 2 million lives. There are numerous ongoing clinical studies aiming to find treatment options and many are being published daily. Some effective treatment options, albeit of variable efficacy, have been discovered. Therefore, it is necessary to develop an evidence-based methodology, to continuously check for new evidence, and to update recommendations accordingly. Here we provide guidelines on pharmaceutical treatment for COVID-19 based on the latest evidence.

Neutralizing antibodies targeted at the receptor-binding domain of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike protein have been developed and now under evaluation in clinical trials. The US Food and Drug Administration currently issued emergency use authorizations for neutralizing monoclonal antibodies in non-hospitalized patients with mild to moderate coronavirus disease 2019 (COVID-19) who are at high risk for progressing to severe disease and/or hospitalization. In terms of this situation, there is an urgent need to investigate the clinical aspects and to develop strategies to deploy them effectively in clinical practice. Here we provide guidance for the use of anti-SARS-CoV-2 monoclonal antibodies for the treatment of COVID-19 based on the latest evidence.