Self-Explainable Graph Neural Network for Alzheimer Disease and Related Dementias Risk Prediction: Algorithm Development and Validation Study.

BACKGROUND: Alzheimer disease and related dementias (ADRD) rank as the sixth leading cause of death in the United States, underlining the importance of accurate ADRD risk prediction. While recent advancements in ADRD risk prediction have primarily relied on imaging analysis, not all patients undergo medical imaging before an ADRD diagnosis. Merging machine learning with claims data can reveal additional risk factors and uncover interconnections among diverse medical codes. OBJECTIVE: The study aims to use graph neural networks (GNNs) with claim data for ADRD risk prediction. Addressing the lack of human-interpretable reasons behind these predictions, we introduce an innovative, self-explainable method to evaluate relationship importance and its influence on ADRD risk prediction. METHODS: We used a variationally regularized encoder-decoder GNN (variational GNN [VGNN]) integrated with our proposed relation importance method for estimating ADRD likelihood. This self-explainable method can provide a feature-important explanation in the context of ADRD risk prediction, leveraging relational information within a graph. Three scenarios with 1-year, 2-year, and 3-year prediction windows were created to assess the model's efficiency, respectively. Random forest (RF) and light gradient boost machine (LGBM) were used as baselines. By using this method, we further clarify the key relationships for ADRD risk prediction. RESULTS: In scenario 1, the VGNN model showed area under the receiver operating characteristic (AUROC) scores of 0.7272 and 0.7480 for the small subset and the matched cohort data set. It outperforms RF and LGBM by 10.6% and 9.1%, respectively, on average. In scenario 2, it achieved AUROC scores of 0.7125 and 0.7281, surpassing the other models by 10.5% and 8.9%, respectively. Similarly, in scenario 3, AUROC scores of 0.7001 and 0.7187 were obtained, exceeding 10.1% and 8.5% than the baseline models, respectively. These results clearly demonstrate the significant superiority of the graph-based approach over the tree-based models (RF and LGBM) in predicting ADRD. Furthermore, the integration of the VGNN model and our relation importance interpretation could provide valuable insight into paired factors that may contribute to or delay ADRD progression. CONCLUSIONS: Using our innovative self-explainable method with claims data enhances ADRD risk prediction and provides insights into the impact of interconnected medical code relationships. This methodology not only enables ADRD risk modeling but also shows potential for other image analysis predictions using claims data.

Prompt Tuning in Biomedical Relation Extraction.

Biomedical relation extraction (RE) is critical in constructing high-quality knowledge graphs and databases as well as supporting many downstream text mining applications. This paper explores prompt tuning on biomedical RE and its few-shot scenarios, aiming to propose a simple yet effective model for this specific task. Prompt tuning reformulates natural language processing (NLP) downstream tasks into masked language problems by embedding specific text prompts into the original input, facilitating the adaption of pre-trained language models (PLMs) to better address these tasks. This study presents a customized prompt tuning model designed explicitly for biomedical RE, including its applicability in few-shot learning contexts. The model's performance was rigorously assessed using the chemical-protein relation (CHEMPROT) dataset from BioCreative VI and the drug-drug interaction (DDI) dataset from SemEval-2013, showcasing its superior performance over conventional fine-tuned PLMs across both datasets, encompassing few-shot scenarios. This observation underscores the effectiveness of prompt tuning in enhancing the capabilities of conventional PLMs, though the extent of enhancement may vary by specific model. Additionally, the model demonstrated a harmonious balance between simplicity and efficiency, matching state-of-the-art performance without needing external knowledge or extra computational resources. The pivotal contribution of our study is the development of a suitably designed prompt tuning model, highlighting prompt tuning's effectiveness in biomedical RE. It offers a robust, efficient approach to the field's challenges and represents a significant advancement in extracting complex relations from biomedical texts. Supplementary Information: The online version contains supplementary material available at 10.1007/s41666-024-00162-9.

Dynamic Prognosis Prediction for Patients on DAPT After Drug-Eluting Stent Implantation: Model Development and Validation.

BACKGROUND: The rapid evolution of artificial intelligence (AI) in conjunction with recent updates in dual antiplatelet therapy (DAPT) management guidelines emphasizes the necessity for innovative models to predict ischemic or bleeding events after drug-eluting stent implantation. Leveraging AI for dynamic prediction has the potential to revolutionize risk stratification and provide personalized decision support for DAPT management. METHODS AND RESULTS: We developed and validated a new AI-based pipeline using retrospective data of drug-eluting stent-treated patients, sourced from the Cerner Health Facts data set (n=98 236) and Optum's de-identified Clinformatics Data Mart Database (n=9978). The 36 months following drug-eluting stent implantation were designated as our primary forecasting interval, further segmented into 6 sequential prediction windows. We evaluated 5 distinct AI algorithms for their precision in predicting ischemic and bleeding risks. Model discriminative accuracy was assessed using the area under the receiver operating characteristic curve, among other metrics. The weighted light gradient boosting machine stood out as the preeminent model, thus earning its place as our AI-DAPT model. The AI-DAPT demonstrated peak accuracy in the 30 to 36 months window, charting an area under the receiver operating characteristic curve of 90% [95% CI, 88%-92%] for ischemia and 84% [95% CI, 82%-87%] for bleeding predictions. CONCLUSIONS: Our AI-DAPT excels in formulating iterative, refined dynamic predictions by assimilating ongoing updates from patients' clinical profiles, holding value as a novel smart clinical tool to facilitate optimal DAPT duration management with high accuracy and adaptability.

Advancing Biomedicine with Graph Representation Learning: Recent Progress, Challenges, and Future Directions.

OBJECTIVES: Graph representation learning (GRL) has emerged as a pivotal field that has contributed significantly to breakthroughs in various fields, including biomedicine. The objective of this survey is to review the latest advancements in GRL methods and their applications in the biomedical field. We also highlight key challenges currently faced by GRL and outline potential directions for future research. METHODS: We conducted a comprehensive search of multiple databases, including PubMed, Web of Science, IEEE Xplore, and Google Scholar, to collect relevant publications from the past two years (2021-2022). The studies selected for review were based on their relevance to the topic and the publication quality. RESULTS: A total of 78 articles were included in our analysis. We identified three main categories of GRL methods and summarized their methodological foundations and notable models. In terms of GRL applications, we focused on two main topics: drug and disease. We analyzed the study frameworks and achievements of the prominent research. Based on the current state-of-the-art, we discussed the challenges and future directions. CONCLUSIONS: GRL methods applied in the biomedical field demonstrated several key characteristics, including the utilization of attention mechanisms to prioritize relevant features, a growing emphasis on model interpretability, and the combination of various techniques to improve model performance. There are also challenges needed to be addressed, including mitigating model bias, accommodating the heterogeneity of large-scale knowledge graphs, and improving the availability of high-quality graph data. To fully leverage the potential of GRL, future efforts should prioritize these areas of research.

Mining on Alzheimer's diseases related knowledge graph to identity potential AD-related semantic triples for drug repurposing.

BACKGROUND: To date, there are no effective treatments for most neurodegenerative diseases. Knowledge graphs can provide comprehensive and semantic representation for heterogeneous data, and have been successfully leveraged in many biomedical applications including drug repurposing. Our objective is to construct a knowledge graph from literature to study the relations between Alzheimer's disease (AD) and chemicals, drugs and dietary supplements in order to identify opportunities to prevent or delay neurodegenerative progression. We collected biomedical annotations and extracted their relations using SemRep via SemMedDB. We used both a BERT-based classifier and rule-based methods during data preprocessing to exclude noise while preserving most AD-related semantic triples. The 1,672,110 filtered triples were used to train with knowledge graph completion algorithms (i.e., TransE, DistMult, and ComplEx) to predict candidates that might be helpful for AD treatment or prevention. RESULTS: Among three knowledge graph completion models, TransE outperformed the other two (MR = 10.53, Hits@1 = 0.28). We leveraged the time-slicing technique to further evaluate the prediction results. We found supporting evidence for most highly ranked candidates predicted by our model which indicates that our approach can inform reliable new knowledge. CONCLUSION: This paper shows that our graph mining model can predict reliable new relationships between AD and other entities (i.e., dietary supplements, chemicals, and drugs). The knowledge graph constructed can facilitate data-driven knowledge discoveries and the generation of novel hypotheses.

A randomized, double-blind, placebo-controlled trial to evaluate the hypoglycemic efficacy of the mcIRBP-19-containing Momordica charantia L. fruit extracts in the type 2 diabetic subjects.

BACKGROUND: The fruits of Momordica charantia L., also named as bitter gourd or bitter melon in popular, is a common tropical vegetable that is traditionally used to reduce blood glucose. A peptide derived from bitter gourd, Momordica charantia insulin receptor binding peptid-19 (mcIRBP-19), had been demonstrated to possess an insulin-like effect in vitro and in the animal studies. However, the benefit of the mcIRBP-19-containing bitter gourd extracts (mcIRBP-19-BGE) for lowering blood glucose levels in humans is unknown. OBJECTIVE: This aim of this study was to evaluate the hypoglycemic efficacy of mcIRBP-19-BGE in subjects with type 2 diabetes who had taken antidiabetic medications but failed to achieve the treatment goal. Whether glucose lowering efficacy of mcIRBP-19-BGE could be demonstrated when the antidiabetic medications were ineffective was also studied. DESIGN: Subjects were randomly assigned to two groups: mcIRBP-19-BGE treatment group (N = 20) and placebo group (N = 20), and were orally administered 600 mg/day investigational product or placebo for 3 months. Subjects whose hemoglobin A1c (HbA1c) continued declining before the trial initiation with the antidiabetic drugs were excluded from the subset analysis to further investigate the efficacy for those who failed to respond to the antidiabetic medications. RESULTS: The oral administration of mcIRBP-19-BGE decreased with a borderline significance at fasting blood glucose (FBG; P = 0.057) and HbA1c (P = 0.060). The subgroup analysis (N = 29) showed that mcIRBP-19-BGE had a significant effect on reducing FBG (from 172.5 ± 32.6 mg/dL to 159.4 ± 18.3 mg/dL, P = 0.041) and HbA1c (from 8.0 ± 0.7% to 7.5 ± 0.8%, P = 0.010). CONCLUSION: All of these results demonstrate that mcIRBP-19-BGE possesses a hypoglycemic effect, and can have a significant reduction in FBG and HbA1c when the antidiabetic drugs are ineffective.

Chemical-Protein Relation Extraction with Pre-trained Prompt Tuning.

Biomedical relation extraction plays a critical role in the construction of high-quality knowledge graphs and databases, which can further support many downstream applications. Pre-trained prompt tuning, as a new paradigm, has shown great potential in many natural language processing (NLP) tasks. Through inserting a piece of text into the original input, prompt converts NLP tasks into masked language problems, which could be better addressed by pre-trained language models (PLMs). In this study, we applied pre-trained prompt tuning to chemical-protein relation extraction using the BioCreative VI CHEMPROT dataset. The experiment results showed that the pre-trained prompt tuning outperformed the baseline approach in chemical-protein interaction classification. We conclude that the prompt tuning can improve the efficiency of the PLMs on chemical-protein relation extraction tasks.

Identification of circular RNAs as novel biomarkers and potentially functional competing endogenous RNA network for myelodysplastic syndrome patients.

Circular RNAs (circRNAs) have been identified to exert vital biological functions and can be used as new biomarkers in a number of tumors. However, little is known about the functions of circRNAs in myelodysplastic syndrome (MDS). Here, we aimed to investigate circRNA expression profiles and to investigate the functional and clinical value of circRNAs in MDS. Differential expression of circRNAs between MDS and control subjects was analyzed using circRNA arrays, in which we identified 145 upregulated circRNAs and 224 downregulated circRNAs. Validated by real-time quantitative PCR between 100 MDS patients and 20 controls, three upregulated (hsa_circRNA_100352, hsa_circRNA_104056, and hsa_circRNA_104634) and three downregulated (hsa_circRNA_103846, hsa_circRNA_102817, and hsa_circRNA_102526) circRNAs matched the arrays. The receiver operating characteristic curve analysis of these circRNAs showed that the area under the curve was 0.7266, 0.8676, 0.7349, 0.7091, 0.8806, and 0.7472, respectively. Kaplan-Meier survival analysis showed that only hsa_circRNA_100352, hsa_circRNA_104056, and hsa_circRNA_102817 were significantly associated with overall survival. Furthermore, we generated a competing endogenous RNA network focused on hsa_circRNA_100352, hsa_circRNA_104056, and hsa_circRNA_102817. Analyses using Gene Ontology and Kyoto Encyclopedia of Genes and Genomes showed that the three circRNAs were linked with some important cancer-related functions and pathways.

Characterization of Peptacetobacter hominis gen. nov., sp. nov., isolated from human faeces, and proposal for the reclassification of Clostridium hiranonis within the genus Peptacetobacter.

A novel, Gram-stain-positive, rod-shaped, non-motile, non-spore-forming, obligately anaerobic bacterium, designated strain ZHW00191T, was isolated from human faeces and characterized by using a polyphasic taxonomic approach. Growth occurred at 25-45 °C (optimum, 37-42 °C), at pH 5.5-10.0 (optimum, pH 6.5-7.0) and with 0-2 % (w/v) NaCl (optimum, 0 %). The end products of glucose fermentation were acetic acid, isobutyric acid and isovaleric acid and a small amount of propionic acid. The dominant cellular fatty acids (>10 %) of strain ZHW00191T were C16 : 0, C18 : 1 ω9с and C18 : 2ω6,9с. Its polar lipid profile comprised diphosphatidylglycerol, phosphatidylglycerol, three unidentified phospholipids and ten unidentified glycolipids. Respiratory quinones were not detected. The cell-wall peptidoglycan contained meso-2,6-diaminopimelic acid, and the whole-cell sugars were ribose and glucose. The genomic DNA G+C content was 32.8 mol%. Analysis of the 16S rRNA gene sequence indicated that ZHW00191T was most closely related to Clostridium hiranonis TO-931T (95.3 % similarity). Average nucleotide identity (ANI) and digital DNA-DNA hybridization (dDDH) analyses with closely related reference strains indicated that reassociation values were both well below the thresholds of 95-96% and 70 % for species delineation, respectively. Based on phenotypic, chemotaxonomic and genetic studies, a novel genus, Peptacetobacter gen. nov., is proposed. The novel isolate ZHW00191T (=JCM 33482T=GDMCC 1.1530T) is proposed as the type strain of the type species Peptacetobacter hominis gen. nov., sp. nov. of the proposed new genus. Furthermore, it is proposed that Clostridium hiranonis be transferred to this novel genus, as Peptacetobacter hiranonis comb. nov.

[Effect on the activities of daily life of the patients with traumatic spinal cord injury treated by the paraplegia-triple-needling method].

OBJECTIVE: To compare the differences in the efficacy on traumatic spinal cord injury(SCI) in the thoracic and lumbar vertebra between the paraplegia-triple-needling method and the conventional acupuncture therapy. METHODS: The perspectively randomized controlled trial was adopted. Forty-eight cases of traumatic SCI in the thoracic and lumbar vertebra were randomized into an observation group and a control group, 24 cases in each one. The conventional rehabilitation training was applied in both groups. In addition, the paraplegia-triple-needling method was used in the observation group. In the treatment, acupuncture was applied to the points of the Governor Vessel and the Back-shu which, located two segments above and below the spinal injury plane separately. Acupuncture with the electric pulsing stimulation was applied to the motor points of the key muscles of the lower extremities. In the control group, the conventional acupuncture was applied to Huantiao (GB 30), Zusanli (ST 36), Xuanzhong (GB 39) and Sanyinjiao (SP 6). In each group, the treatment was given once a day, one month treatment made 1 session. Totally, 3 sessions of treatment were required. Before and after treatment, as well as in 1-month follow-up visit after treatment, the modified Barthel index (MBI) and the function comprehensive assessment (FCA) were adopted to assess the activities of daily life (ADL) and the comprehensive function of the patients. The score of MBI and FCA were taken as the double response variables to imitate the multilevel model. The changing tendency of MBI and FCA along with the time was observed in two groups. RESULTS: In the follow-up visit, MBI and FCA score were all improved as compared with those before treatment in two groups (all P < 0.05). There were no statistically significant differences in MBI and FCA score at any time point between two groups (all P > 0.05). In 4-month observation, there was a rising tendency with time in MBI and FAC scoe in both groups, which was roughly linear. As time went on, the increasing amplitude in the observation group was much bigger. It was explained that there was no difference in the short-term efficacy between two groups. However, the long-term efficacy in the observation group was much better. CONCLUSION: Both the paraplegia-triple-needling method and the conventional acupuncture therapy can improve the ADL and the comprehensive function of the patients with traumatic SCI of the thoracic and lumbar vertebra. Concerning the long-term efficacy, the paraplegia-triple-needling combined with the rehabilitation training achieves better result. This therapeutic program is safe and effective.

[Spectroscopic studies of aripiprazole].

The ultraviolet spectrum (UV), infrared spectrum (IR), nuclear magnetic resonance (NMR) and mass spectrum (MS) of aripiprazole, a new antipsychotic drug, were reported and interpreted. The structure of aripiprazole in solution was studied according to the UV spectra detected in solution with different pH values. The vibrations of functional groups of this compound in IR and the isotopic ion peaks in MS were discussed. Moreover, the 2D-NMR techniques, including 1H-1H correlation spectroscopy (1H-1H cosy), heteronuclear single-quantum coherence (HSQC), and heteronuclear multiple-bond correlation (HMBC), were used to deduce the structure of this compound. All the 1H NMR and 13C NMR signals were assigned. Especially, the ten different methylenes in this structure were analyzed according to the chemical shifts, coupling constants and correlations in 2D-NMR spectrum. By all these spectral techniques, the structure of aripiprazole was identified.

Using support vector classification for SAR of fentanyl derivatives.

AIM: To discriminate between fentanyl derivatives with high and low activities. METHODS: The support vector classification (SVC) method, a novel approach, was employed to investigate structure-activity relationship (SAR) of fentanyl derivatives based on the molecular descriptors, which were quantum parameters including DeltaE [energy difference between highest occupied molecular orbital energy (HOMO) and lowest empty molecular orbital energy (LUMO)], MR (molecular refractivity) and M(r) (molecular weight). RESULTS: By using leave-one-out cross-validation test, the accuracies of prediction for activities of fentanyl derivatives in SVC, principal component analysis (PCA), artificial neural network (ANN) and K-nearest neighbor (KNN) models were 93%, 86%, 57%, and 71%, respectively. The results indicated that the performance of the SVC model was better than those of PCA, ANN, and KNN models for this data. CONCLUSION: SVC can be used to investigate SAR of fentanyl derivatives and could be a promising tool in the field of SAR research.

Semiempirical quantum chemical method and artificial neural networks applied for lambdamax computation of some azo dyes.

The maximum absorption wavelengths of 31 azo dyes have been calculated by two comprehensive methods using the semiempirical quantum chemical method, PM3, and the weight decay based artificial neural network (WD-ANN) or the early stopping based artificial neural network (ES-ANN). The average absolute errors of WD-ANN and that of ES-ANN are 10.07 nm and 12.40 nm, respectively. These results are much better than the results using ZINDO/S with the default value (0.585) only.

Hyper-polyhedron model applied to molecular screening of guanidines as Na/H exchange inhibitors.

AIM: To investigate structure-activity relationships of N-(3-Oxo-3,4-dihydro-2H-benzo[1,4]oxazine-6-carbonyl) guanidines in Na/H exchange inhibitory activities and probe into a new method of the computer-aided molecular screening. METHODS: The hyper-polyhedron model (HPM) was proposed in our lab. RESULTS: The samples with probably higher activities could be determined in such a way that their representing points should be in the hyper-polyhedron region where all known samples with high activities were distributed. And the predictive ability of different methods available was tested by the cross-validation experiment. CONCLUSION: The accurate rate of molecular screening of N-(3-Oxo-3,4-dihydro-2H-benzo[1,4]oxazine-6-carbonyl) guanidines by HPM was much higher than that obtained by PCA (principal component analysis) and Fisher methods for the data set available here. Therefore, HPM could be used as a powerful tool for screening new compounds with probably higher activities.