An efficient curriculum learning-based strategy for molecular graph learning.

Computational methods have been widely applied to resolve various core issues in drug discovery, such as molecular property prediction. In recent years, a data-driven computational method-deep learning had achieved a number of impressive successes in various domains. In drug discovery, graph neural networks (GNNs) take molecular graph data as input and learn graph-level representations in non-Euclidean space. An enormous amount of well-performed GNNs have been proposed for molecular graph learning. Meanwhile, efficient use of molecular data during training process, however, has not been paid enough attention. Curriculum learning (CL) is proposed as a training strategy by rearranging training queue based on calculated samples' difficulties, yet the effectiveness of CL method has not been determined in molecular graph learning. In this study, inspired by chemical domain knowledge and task prior information, we proposed a novel CL-based training strategy to improve the training efficiency of molecular graph learning, called CurrMG. Consisting of a difficulty measurer and a training scheduler, CurrMG is designed as a plug-and-play module, which is model-independent and easy-to-use on molecular data. Extensive experiments demonstrated that molecular graph learning models could benefit from CurrMG and gain noticeable improvement on five GNN models and eight molecular property prediction tasks (overall improvement is 4.08%). We further observed CurrMG's encouraging potential in resource-constrained molecular property prediction. These results indicate that CurrMG can be used as a reliable and efficient training strategy for molecular graph learning. Availability: The source code is available in https://github.com/gu-yaowen/CurrMG.

Prediction of Medical Disputes Between Health Care Workers and Patients in Terms of Hospital Legal Construction Using Machine Learning Techniques: Externally Validated Cross-Sectional Study.

BACKGROUND: Medical disputes are a global public health issue that is receiving increasing attention. However, studies investigating the relationship between hospital legal construction and medical disputes are scarce. The development of a multicenter model incorporating machine learning (ML) techniques for the individualized prediction of medical disputes would be beneficial for medical workers. OBJECTIVE: This study aimed to identify predictors related to medical disputes from the perspective of hospital legal construction and the use of ML techniques to build models for predicting the risk of medical disputes. METHODS: This study enrolled 38,053 medical workers from 130 tertiary hospitals in Hunan province, China. The participants were randomly divided into a training cohort (34,286/38,053, 90.1%) and an internal validation cohort (3767/38,053, 9.9%). Medical workers from 87 tertiary hospitals in Beijing were included in an external validation cohort (26,285/26,285, 100%). This study used logistic regression and 5 ML techniques: decision tree, random forest, support vector machine, gradient boosting decision tree (GBDT), and deep neural network. In total, 12 metrics, including discrimination and calibration, were used for performance evaluation. A scoring system was developed to select the optimal model. Shapley additive explanations was used to generate the importance coefficients for characteristics. To promote the clinical practice of our proposed optimal model, reclassification of patients was performed, and a web-based app for medical dispute prediction was created, which can be easily accessed by the public. RESULTS: Medical disputes occurred among 46.06% (17,527/38,053) of the medical workers in Hunan province, China. Among the 26 clinical characteristics, multivariate analysis demonstrated that 18 characteristics were significantly associated with medical disputes, and these characteristics were used for ML model development. Among the ML techniques, GBDT was identified as the optimal model, demonstrating the lowest Brier score (0.205), highest area under the receiver operating characteristic curve (0.738, 95% CI 0.722-0.754), and the largest discrimination slope (0.172) and Youden index (1.355). In addition, it achieved the highest metrics score (63 points), followed by deep neural network (46 points) and random forest (45 points), in the internal validation set. In the external validation set, GBDT still performed comparably, achieving the second highest metrics score (52 points). The high-risk group had more than twice the odds of experiencing medical disputes compared with the low-risk group. CONCLUSIONS: We established a prediction model to stratify medical workers into different risk groups for encountering medical disputes. Among the 5 ML models, GBDT demonstrated the optimal comprehensive performance and was used to construct the web-based app. Our proposed model can serve as a useful tool for identifying medical workers at high risk of medical disputes. We believe that preventive strategies should be implemented for the high-risk group.

Employing Molecular Conformations for Ligand-Based Virtual Screening with Equivariant Graph Neural Network and Deep Multiple Instance Learning.

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.

Comparison of Mortality Predictive Models of Sepsis Patients Based on Machine Learning.

Objective To compare the performance of five machine learning models and SAPS II score in predicting the 30-day mortality amongst patients with sepsis. Methods The sepsis patient-related data were extracted from the MIMIC-IV database. Clinical features were generated and selected by mutual information and grid search. Logistic regression, Random forest, LightGBM, XGBoost, and other machine learning models were constructed to predict the mortality probability. Five measurements including accuracy, precision, recall, F1 score, and area under curve (AUC) were acquired for model evaluation. An external validation was implemented to avoid conclusion bias. Results LightGBM outperformed other methods, achieving the highest AUC (0.900), accuracy (0.808), and precision (0.559). All machine learning models performed better than SAPS II score (AUC=0.748). LightGBM achieved 0.883 in AUC in the external data validation. Conclusions The machine learning models are more effective in predicting the 30-day mortality of patients with sepsis than the traditional SAPS II score.

Developing a Personalized Meal Recommendation System for Chinese Older Adults: Observational Cohort Study.

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.

MGREL: A multi-graph representation learning-based ensemble learning method for gene-disease association prediction.

The identification of gene-disease associations plays an important role in the exploration of pathogenic mechanisms and therapeutic targets. Computational methods have been regarded as an effective way to discover the potential gene-disease associations in recent years. However, most of them ignored the combination of abundant genetic, therapeutic information, and gene-disease network topology. To this end, we re-organized the current gene-disease association benchmark dataset by extracting the newest gene-disease associations from the OMIM database. Then, we developed a multi-graph representation learning-based ensemble model, named MGREL to predict gene-disease associations. MGREL integrated two feature generation channels to extract gene and disease features, including a knowledge extraction channel which learned high-order representations from genetic and therapeutic information, and a graph learning channel which acquired network topological representations through multiple advanced graph representation learning methods. Then, an ensemble learning method with 5 machine learning models was used as the classifier to predict the gene-disease association. Comprehensive experiments have demonstrated the significant performance achieved by MGREL compared to 5 state-of-the-art methods. For the major measurements (AUC = 0.925, AUPR = 0.935), the relative improvements of MGREL compared to the suboptimal methods are 3.24%, and 2.75%, respectively. MGREL also achieved impressive improvements in the challenging tasks of predicting potential associations for unknown genes/diseases. In addition, case studies implied potential applications for MGREL in the discovery of potential therapeutic targets.

Drug-disease association prediction with literature based multi-feature fusion.

Introduction: Exploring the potential efficacy of a drug is a valid approach for drug development with shorter development times and lower costs. Recently, several computational drug repositioning methods have been introduced to learn multi-features for potential association prediction. However, fully leveraging the vast amount of information in the scientific literature to enhance drug-disease association prediction is a great challenge. Methods: We constructed a drug-disease association prediction method called Literature Based Multi-Feature Fusion (LBMFF), which effectively integrated known drugs, diseases, side effects and target associations from public databases as well as literature semantic features. Specifically, a pre-training and fine-tuning BERT model was introduced to extract literature semantic information for similarity assessment. Then, we revealed drug and disease embeddings from the constructed fusion similarity matrix by a graph convolutional network with an attention mechanism. Results: LBMFF achieved superior performance in drug-disease association prediction with an AUC value of 0.8818 and an AUPR value of 0.5916. Discussion: LBMFF achieved relative improvements of 31.67% and 16.09%, respectively, over the second-best results, compared to single feature methods and seven existing state-of-the-art prediction methods on the same test datasets. Meanwhile, case studies have verified that LBMFF can discover new associations to accelerate drug development. The proposed benchmark dataset and source code are available at: https://github.com/kang-hongyu/LBMFF.

Study on tumour cell-derived hybrid exosomes as dasatinib nanocarriers for pancreatic cancer therapy.

Pancreatic ductal adenocarcinoma (PDAC) is one of the leading causes of cancer-related death. Therefore, we intend to explore novel strategies against PDAC. The exosomes-based biomimetic nanoparticle is an appealing candidate served as a drug carrier in cancer treatment, due to its inherit abilities. In the present study, we designed dasatinib-loaded hybrid exosomes by fusing human pancreatic cancer cells derived exosomes with dasatinib-loaded liposomes, followed by characterization for particle size (119.9 ± 6.10 nm) and zeta potential (-11.45 ± 2.24 mV). Major protein analysis from western blot techniques reveal the presence of exosome marker proteins CD9 and CD81. PEGylated hybrid exosomes showed pH-sensitive drug release in acidic condition, benefiting drug delivery to acidic cancer environment. Dasatinib-loaded hybrid exosomes exhibited significantly higher uptake rates and cytotoxicity to parent PDAC cells by two-sample t-test or by one-way ANOVA analysis of variance, as compared to free drug or liposomal formulations. The results from our computational analysis demonstrated that the drug-likeness, ADMET, and protein-ligand binding affinity of dasatinib are verified successfully. Cancer derived hybrid exosomes may serve as a potential therapeutic candidate for pancreatic cancer treatment.

REDDA: Integrating multiple biological relations to heterogeneous graph neural network for drug-disease association prediction.

Computational drug repositioning is an effective way to find new indications for existing drugs, thus can accelerate drug development and reduce experimental costs. Recently, various deep learning-based repurposing methods have been established to identify the potential drug-disease associations (DDA). However, effective utilization of the relations of biological entities to capture the biological interactions to enhance the drug-disease association prediction is still challenging. To resolve the above problem, we proposed a heterogeneous graph neural network called REDDA (Relations-Enhanced Drug-Disease Association prediction). Assembled with three attention mechanisms, REDDA can sequentially learn drug/disease representations by a general heterogeneous graph convolutional network-based node embedding block, a topological subnet embedding block, a graph attention block, and a layer attention block. Performance comparisons on our proposed benchmark dataset show that REDDA outperforms 8 advanced drug-disease association prediction methods, achieving relative improvements of 0.76% on the area under the receiver operating characteristic curve (AUC) score and 13.92% on the precision-recall curve (AUPR) score compared to the suboptimal method. On the other benchmark dataset, REDDA also obtains relative improvements of 2.48% on the AUC score and 4.93% on the AUPR score. Specifically, case studies also indicate that REDDA can give valid predictions for the discovery of -new indications for drugs and new therapies for diseases. The overall results provide an inspiring potential for REDDA in the in silico drug development. The proposed benchmark dataset and source code are available in https://github.com/gu-yaowen/REDDA.

Developing a Siamese Network for UTIs Risk Prediction in Immobile Patients Undergoing Stroke.

Stroke patients tend to suffer from immobility, which increases the possibility of post-stroke complications. Urinary tract infections (UTIs) are one of the complications as an independent predictor of poor prognosis of stroke patients. However, the incidence of new UTIs onsets during hospitalization was rare in most datasets with a prevalence of 4%. This imbalanced data distribution sets obstacles to establishing an accurate prediction model. Our study aimed to develop an effective prediction model to identify UTIs risk in immobile stroke patients, and (2) to compare its prediction performance with traditional machine learning models. We tackled this problem by building a Siamese Network leveraging commonly used clinical features to identifying patients with UTIs risk. Model derivation and validation were based on a nationwide dataset including 3982 Chinese patients. Results showed that the Siamese Network performed better than traditional machine learning models in imbalanced datasets (Sensitivity: 0.810; AUC: 0.828).