Hongyu Kang

The etiology of pediatric costochondritis, a common cause of musculoskeletal chest pain, remains poorly understood. This study aimed to identify nutritiona

Background: Drug repositioning is a pivotal strategy in pharmaceutical research, offering accelerated and cost-effective therapeutic discovery. However, biomedical information relevant to drug repositioning is often complex, dispersed, and underutilized due to limitations in traditional extraction methods, such as reliance on annotated data and poor generalizability. Large language models (LLMs) show promise but face challenges such as hallucinations and interpretability issues. Objective: This study proposed long chain-of-thought for drug repositioning knowledge extraction (LCoDR-KE), a lightweight and domain-specific framework to enhance LLMs’ accuracy and adaptability in extracting structured biomedical knowledge for drug repositioning. Methods: A domain-specific schema defined 11 entities (eg, drug, disease) and 18 relationships (eg, treats, is biomarker of). Following the established schema architecture, we constructed automatic annotation based on 10,000 PubMed abstracts via chain-of-thought prompt engineering. A total of 1000 expert-validated abstracts were curated into a drug repositioning corpus, a high-quality specialized corpus, while the remaining entries were allocated for model training purposes. Then, the proposed LCoDR-KE framework combined supervised fine-tuning of the Qwen2.5-7B-Instruct model with reinforcement learning and dual-reward mechanisms. Performance was evaluated against state-of-the-art models (eg, conditional random fields, Bidirectional Encoder Representations From Transformers, BioBERT, Qwen2.5, DeepSeek-R1, OpenBioLLM-70B, and model variants) using precision, recall, and F₁-score. In addition, the convergence of the training method was assessed by analyzing performance progression across iteration steps. Results: LCoDR-KE achieved an entity F₁ of 81.46% (eg, drug 95.83%, disease 90.52%) and triplet F₁ of 69.04%, outperforming traditional models and rivaling larger LLMs (DeepSeek-R1: entity F₁=84.64%, triplet F₁=69.02%). Ablation studies confirmed the contributions of supervised fine-tuning (8.61% and 20.70% F₁ drop if removed) and reinforcement learning (6.09% and 14.09% F₁ drop if removed). The training process demonstrated stable convergence, validated through iterative performance monitoring. Qualitative analysis of the model’s chain-of-thought outputs showed that LCoDR-KE performed structured and schema-aware reasoning by validating entity types, rejecting incompatible relations, enforcing constraints, and generating compliant JSON. Error analysis revealed 4 main types of mistakes and challenges for further improvement. Conclusions: LCoDR-KE enhances LLMs’ domain-specific adaptability for drug repositioning by offering an open-source drug repositioning corpus and a long chain-of-thought framework based on a lightweight LLM model. This framework supports drug discovery and knowledge reasoning while providing scalable, interpretable solutions applicable to broader biomedical knowledge extraction tasks.

Background: Nutrient needs vary over the lifespan. Improving knowledge of both population groups and care providers can help with healthier food choices, thereby promoting population health and preventing diseases. Providing evidence-based food knowledge online is credible, low cost, and easily accessible. Objective: This study aimed to develop an online multimodal food data exploration platform for easy access to evidence-based diet- and nutrition-related data. Methods: We developed an online platform named Food Atlas in collaboration with a multidisciplinary expert group from the National Institute for Nutrition and Health and Peking Union Medical College Hospital in China. To demonstrate its feasibility for Chinese food for pregnant women, a user-friendly and high-quality multimodal food knowledge graph was constructed, and various interactions with graph-structured data were developed for easy access, including graph-based interactive visualizations, natural language retrieval, and image-text retrieval. Subsequently, we evaluated Food Atlas from both the system perspective and the user perspective. Results: The constructed multimodal food knowledge graph contained a total of 2011 entities, 10,410 triplets, and 23,497 images. Its schema consisted of 11 entity types and 26 types of semantic relations. Compared with 5 other online dietary platforms (Foodwake, Boohee, Xiachufang, Allrecipes, and Yummly), Food Atlas offers a distinct and comprehensive set of data content and system functions desired by target populations. Meanwhile, a total of 28 participants representing 4 different user groups were recruited to evaluate its usability: preparing for pregnancy (n=8), pregnant (n=12), clinicians (n=5), and dietitians (n=3). The mean System Usability Scale index of our platform was 82.5 (SD 9.94; range 40.0-82.5). This above-average usability score and the use cases indicated that Food Atlas is tailored to the needs of the target users. Furthermore, 96% (27/28) of the participants stated that the platform had high consistency, illustrating the necessity and effectiveness of health professionals participating in online, evidence-based resource development. Conclusions: This study demonstrates the development of an online multimodal food data exploration platform and its ability to meet the rising demand for accessible, credible, and appropriate evidence-based online dietary resources. Further research and broader implementation of such platforms have the potential to popularize knowledge, thereby helping populations at different life stages make healthier food choices.

Background: China’s older population is facing serious health challenges, including malnutrition and multiple chronic conditions. There is a critical need for tailored food recommendation systems. Knowledge graph–based food recommendations offer considerable promise in delivering personalized nutritional support. However, the integration of disease-based nutritional principles and preference-related requirements needs to be optimized in current recommendation processes. Objective: This study aims to develop a knowledge graph–based personalized meal recommendation system for community-dwelling older adults and to conduct preliminary effectiveness testing. Methods: We developed ElCombo, a personalized meal recommendation system driven by user profiles and food knowledge graphs. User profiles were established from a survey of 96 community-dwelling older adults. Food knowledge graphs were supported by data from websites of Chinese cuisine recipes and eating history, consisting of 5 entity classes: dishes, ingredients, category of ingredients, nutrients, and diseases, along with their attributes and interrelations. A personalized meal recommendation algorithm was then developed to synthesize this information to generate packaged meals as outputs, considering disease-related nutritional constraints and personal dietary preferences. Furthermore, a validation study using a real-world data set collected from 96 community-dwelling older adults was conducted to assess ElCombo’s effectiveness in modifying their dietary habits over a 1-month intervention, using simulated data for impact analysis. Results: Our recommendation system, ElCombo, was evaluated by comparing the dietary diversity and diet quality of its recommended meals with those of the autonomous choices of 96 eligible community-dwelling older adults. Participants were grouped based on whether they had a recorded eating history, with 34 (35%) having and 62 (65%) lacking such data. Simulation experiments based on retrospective data over a 30-day evaluation revealed that ElCombo’s meal recommendations consistently had significantly higher diet quality and dietary diversity compared to the older adults’ own selections (P<.001). In addition, case studies of 2 older adults, 1 with and 1 without prior eating records, showcased ElCombo’s ability to fulfill complex nutritional requirements associated with multiple morbidities, personalized to each individual’s health profile and dietary requirements. Conclusions: ElCombo has shown enhanced potential for improving dietary quality and diversity among community-dwelling older adults in simulation tests. The evaluation metrics suggest that the food choices supported by the personalized meal recommendation system surpass autonomous selections. Future research will focus on validating and refining ElCombo’s performance in real-world settings, emphasizing the robust management of complex health data. The system’s scalability and adaptability pinpoint its potential for making a meaningful impact on the nutritional health of older adults.

Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet #Preprint #PeerReviewMe: Warning: This is a unreviewed preprint. Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn. Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period. Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author). Background: Nutrient needs vary over the lifespan. Improving knowledge of both population groups and care providers can help with healthier food choices, thereby promoting population health and preventing diseases. Providing evidence-based food knowledge online is credible, low cost, and easily accessible. Objective: This study aimed to develop an online multimodal food data exploration platform for easy access to evidence-based diet- and nutrition-related data. Methods: We developed an online platform named Food Atlas in collaboration with a multidisciplinary expert group from the National Institute for Nutrition and Health and Peking Union Medical College Hospital in China. To demonstrate its feasibility for Chinese food for pregnant women, a user-friendly and high-quality multimodal food knowledge graph was constructed, and various interactions with graph-structured data were developed for easy access, including graph-based interactive visualizations, natural language retrieval, and image-text retrieval. Subsequently, we evaluated Food Atlas from both the system perspective and the user perspective. Results: The constructed multimodal food knowledge graph contained a total of 2011 entities, 10,410 triplets, and 23,497 images. Its schema consisted of 11 entity types and 26 types of semantic relations. Compared with 5 other online dietary platforms (Foodwake, Boohee, Xiachufang, Allrecipes, and Yummly), Food Atlas offers a distinct and comprehensive set of data content and system functions desired by target populations. Meanwhile, a total of 28 participants representing 4 different user groups were recruited to evaluate its usability: preparing for pregnancy (n=8), pregnant (n=12), clinicians (n=5), and dietitians (n=3). The mean System Usability Scale index of our platform was 82.5 (SD 9.94; range 40.0-82.5). This above-average usability score and the use cases indicated that Food Atlas is tailored to the needs of the target users. Furthermore, 96% (27/28) of the participants stated that the platform had high consistency, illustrating the necessity and effectiveness of health professionals participating in online, evidence-based resource development. Conclusions: This study demonstrates the development of an online multimodal food data exploration platform and its ability to meet the rising demand for accessible, credible, and appropriate evidence-based online dietary resources. Further research and broader implementation of such platforms have the potential to popularize knowledge, thereby helping populations at different life stages make healthier food choices.

BACKGROUND China’s older population is facing serious health challenges, including malnutrition and multiple chronic conditions. There is a critical need for tailored food recommendation systems. Knowledge graph–based food recommendations offer considerable promise in delivering personalized nutritional support. However, the integration of disease-based nutritional principles and preference-related requirements needs to be optimized in current recommendation processes. OBJECTIVE This study aims to develop a knowledge graph–based personalized meal recommendation system for community-dwelling older adults and to conduct preliminary effectiveness testing. METHODS We developed ElCombo, a personalized meal recommendation system driven by user profiles and food knowledge graphs. User profiles were established from a survey of 96 community-dwelling older adults. Food knowledge graphs were supported by data from websites of Chinese cuisine recipes and eating history, consisting of 5 entity classes: dishes, ingredients, category of ingredients, nutrients, and diseases, along with their attributes and interrelations. A personalized meal recommendation algorithm was then developed to synthesize this information to generate packaged meals as outputs, considering disease-related nutritional constraints and personal dietary preferences. Furthermore, a validation study using a real-world data set collected from 96 community-dwelling older adults was conducted to assess ElCombo’s effectiveness in modifying their dietary habits over a 1-month intervention, using simulated data for impact analysis. RESULTS Our recommendation system, ElCombo, was evaluated by comparing the dietary diversity and diet quality of its recommended meals with those of the autonomous choices of 96 eligible community-dwelling older adults. Participants were grouped based on whether they had a recorded eating history, with 34 (35%) having and 62 (65%) lacking such data. Simulation experiments based on retrospective data over a 30-day evaluation revealed that ElCombo’s meal recommendations consistently had significantly higher diet quality and dietary diversity compared to the older adults’ own selections (P<.001). In addition, case studies of 2 older adults, 1 with and 1 without prior eating records, showcased ElCombo’s ability to fulfill complex nutritional requirements associated with multiple morbidities, personalized to each individual’s health profile and dietary requirements. CONCLUSIONS ElCombo has shown enhanced potential for improving dietary quality and diversity among community-dwelling older adults in simulation tests. The evaluation metrics suggest that the food choices supported by the personalized meal recommendation system surpass autonomous selections. Future research will focus on validating and refining ElCombo’s performance in real-world settings, emphasizing the robust management of complex health data. The system’s scalability and adaptability pinpoint its potential for making a meaningful impact on the nutritional health of older adults.

Ligand-based virtual screening (LBVS) is a promising approach for rapid and low-cost screening of potentially bioactive molecules in the early stage of drug discovery. Compared with traditional similarity-based machine learning methods, deep learning frameworks for LBVS can more effectively extract high-order molecule structure representations from molecular fingerprints or structures. However, the 3D conformation of a molecule largely influences its bioactivity and physical properties, and has rarely been considered in previous deep learning-based LBVS methods. Moreover, the relative bioactivity benchmark dataset is still lacking. To address these issues, we introduce a novel end-to-end deep learning architecture trained from molecular conformers for LBVS. We first extracted molecule conformers from multiple public molecular bioactivity data and consolidated them into a large-scale bioactivity benchmark dataset, which totally includes millions of endpoints and molecules corresponding to 954 targets. Then, we devised a deep learning-based LBVS called EquiVS to learn molecule representations from conformers for bioactivity prediction. Specifically, graph convolutional network (GCN) and equivariant graph neural network (EGNN) are sequentially stacked to learn high-order molecule-level and conformer-level representations, followed with attention-based deep multiple-instance learning (MIL) to aggregate these representations and then predict the potential bioactivity for the query molecule on a given target. We conducted various experiments to validate the data quality of our benchmark dataset, and confirmed EquiVS achieved better performance compared with 10 traditional machine learning or deep learning-based LBVS methods. Further ablation studies demonstrate the significant contribution of molecular conformation for bioactivity prediction, as well as the reasonability and non-redundancy of deep learning architecture in EquiVS. Finally, a model interpretation case study on CDK2 shows the potential of EquiVS in optimal conformer discovery. The overall study shows that our proposed benchmark dataset and EquiVS method have promising prospects in virtual screening applications.

Ligand-based virtual screening (LBVS) is a promising approach for rapid and low-cost screening of potentially bioactive molecules in the early stage of drug discovery. Compared with traditional similarity-based machine learning methods, deep learning frameworks for LBVS can more effectively extract high-order molecule structure representations from molecular fingerprints or structures. However, the 3D conformation of a molecule largely influences its bioactivity and physical properties, and has rarely been considered in previous deep learning-based LBVS methods. Moreover, the relative bioactivity benchmark dataset is still lacking. To address these issues, we introduce a novel end-to-end deep learning architecture trained from molecular conformers for LBVS. We first extracted molecule conformers from multiple public molecular bioactivity data and consolidated them into a large-scale bioactivity benchmark dataset, which totally includes millions of endpoints and molecules corresponding to 954 targets. Then, we devised a deep learning-based LBVS called EquiVS to learn molecule representations from conformers for bioactivity prediction. Specifically, graph convolutional network (GCN) and equivariant graph neural network (EGNN) are sequentially stacked to learn high-order molecule-level and conformer-level representations, followed by attention-based deep multiple-instance learning (MIL) to aggregate these representations and then predict the potential bioactivity for the query molecule on a given target. We conducted various experiments to validate the data quality of our benchmark dataset, and confirmed EquiVS achieved better performance compared with 10 traditional machine learning or deep learning-based LBVS methods. Further ablation studies demonstrate the significant contribution of molecular conformation for bioactivity prediction, as well as the reasonability and non-redundancy of deep learning architecture in EquiVS. Finally, a model interpretation case study on CDK2 shows the potential of EquiVS in optimal conformer discovery. The overall study shows that our proposed benchmark dataset and EquiVS method have promising prospects in virtual screening applications.

Background: Many drugs do not work the same way for everyone owing to distinctions in their genes. Pharmacogenomics (PGx) aims to understand how genetic variants influence drug efficacy and toxicity. It is often considered one of the most actionable areas of the personalized medicine paradigm. However, little prior work has included in-depth explorations and descriptions of drug usage, dosage adjustment, and so on. Objective: We present a pharmacogenomics knowledge model to discover the hidden relationships between PGx entities such as drugs, genes, and diseases, especially details in precise medication. Methods: PGx open data such as DrugBank and RxNorm were integrated in this study, as well as drug labels published by the US Food and Drug Administration. We annotated 190 drug labels manually for entities and relationships. Based on the annotation results, we trained 3 different natural language processing models to complete entity recognition. Finally, the pharmacogenomics knowledge model was described in detail. Results: In entity recognition tasks, the Bidirectional Encoder Representations from Transformers–conditional random field model achieved better performance with micro-F1 score of 85.12%. The pharmacogenomics knowledge model in our study included 5 semantic types: drug, gene, disease, precise medication (population, daily dose, dose form, frequency, etc), and adverse reaction. Meanwhile, 26 semantic relationships were defined in detail. Taking melanoma caused by a BRAF gene mutation into consideration, the pharmacogenomics knowledge model covered 7 related drugs and 4846 triples were established in this case. All the corpora, relationship definitions, and triples were made publically available. Conclusions: We highlighted the pharmacogenomics knowledge model as a scalable framework for clinicians and clinical pharmacists to adjust drug dosage according to patient-specific genetic variation, and for pharmaceutical researchers to develop new drugs. In the future, a series of other antitumor drugs and automatic relation extractions will be taken into consideration to further enhance our framework with more PGx linked data.

BACKGROUND Many drugs do not work the same way for everyone owing to distinctions in their genes. Pharmacogenomics (PGx) aims to understand how genetic variants influence drug efficacy and toxicity. It is often considered one of the most actionable areas of the personalized medicine paradigm. However, little prior work has included in-depth explorations and descriptions of drug usage, dosage adjustment, and so on. OBJECTIVE We present a pharmacogenomics knowledge model to discover the hidden relationships between PGx entities such as drugs, genes, and diseases, especially details in precise medication. METHODS PGx open data such as DrugBank and RxNorm were integrated in this study, as well as drug labels published by the US Food and Drug Administration. We annotated 190 drug labels manually for entities and relationships. Based on the annotation results, we trained 3 different natural language processing models to complete entity recognition. Finally, the pharmacogenomics knowledge model was described in detail. RESULTS In entity recognition tasks, the Bidirectional Encoder Representations from Transformers–conditional random field model achieved better performance with micro-F1 score of 85.12%. The pharmacogenomics knowledge model in our study included 5 semantic types: drug, gene, disease, precise medication (population, daily dose, dose form, frequency, etc), and adverse reaction. Meanwhile, 26 semantic relationships were defined in detail. Taking melanoma caused by a BRAF gene mutation into consideration, the pharmacogenomics knowledge model covered 7 related drugs and 4846 triples were established in this case. All the corpora, relationship definitions, and triples were made publically available. CONCLUSIONS We highlighted the pharmacogenomics knowledge model as a scalable framework for clinicians and clinical pharmacists to adjust drug dosage according to patient-specific genetic variation, and for pharmaceutical researchers to develop new drugs. In the future, a series of other antitumor drugs and automatic relation extractions will be taken into consideration to further enhance our framework with more PGx linked data.