Artificial Intelligence for Data Discovery and Reuse in Endocrinology and Metabolism

In an intermediate integration strategy, multi-omics features and mechanistic model-derived features can be used for an independent phase of modality-specific training, before being fused in a single set of latent features. These can then be used within existing ML architectures (e.g. a fully connected neural network) to achieve the final prediction.

Editorial

05 May 2023

Editorial: Artificial intelligence for data discovery and reuse in endocrinology and metabolism

Claudio Angione

Huajin Wang

and

Noël Burtt

3,449 views

0 citations

Editors

Huajin Wang

Carnegie Mellon University

Impact

Original Research

20 January 2023

Implementation of five machine learning methods to predict the 52-week blood glucose level in patients with type 2 diabetes

Xiaomin Fu

, 7 more and

Shuangtong Yan

Objective: For the patients who are suffering from type 2 diabetes, blood glucose level could be affected by multiple factors. An accurate estimation of the trajectory of blood glucose is crucial in clinical decision making. Frequent glucose measurement serves as a good source of data to train machine learning models for prediction purposes. This study aimed at using machine learning methods to predict blood glucose for type 2 diabetic patients. We investigated various parameters influencing blood glucose, as well as determined the most effective machine learning algorithm in predicting blood glucose.

Patients and methods: 273 patients were recruited in this research. Several parameters such as age, diet, family history, BMI, alcohol intake, smoking status et al were analyzed. Patients who had glycosylated hemoglobin less than 6.5% after 52 weeks were considered as having achieved glycemic control and the rest as not achieving it. Five machine learning methods (KNN algorithm, logistic regression algorithm, random forest algorithm, support vector machine, and XGBoost algorithm) were compared to evaluate their performances in prediction accuracy. R 3.6.3 and Python 3.12 were used in data analysis.

Results: The statistical variables for which p< 0.05 was obtained were BMI, pulse, Na, Cl, AKP. Compared with the other four algorithms, XGBoost algorithm has the highest accuracy (Accuracy=99.54% in training set and 78.18% in testing set) and AUC values (1.0 in training set and 0.68 in testing set), thus it is recommended to be used for prediction in clinical practice.

Conclusion: When it comes to future blood glucose level prediction using machine learning methods, XGBoost algorithm scores the highest in effectiveness. This algorithm could be applied to assist clinical decision making, as well as guide the lifestyle of diabetic patients, in pursuit of minimizing risks of hyperglycemic or hypoglycemic events.

4,002 views

7 citations

P<0.05 was checked by Pearson’s correlation coefficient. RNA-seq, RNA sequencing; qRT-PCR, quantitative real-time polymerase chain reaction; DEGs, differentially expressed genes.

Original Research

04 January 2023

Transcriptomic profiling of hepatic tissues for drug metabolism genes in nonalcoholic fatty liver disease: A study of human and animals

Li Chen

, 7 more and

Yu-He Wang

3,708 views

3 citations

Original Research

19 October 2022

Combining bioinformatics, network pharmacology and artificial intelligence to predict the mechanism of celastrol in the treatment of type 2 diabetes

Ming Wu

and

Yan Zhang

Heatmap display of the clustering of differentially expressed genes (DEGs). Red indicates the group of healthy human blood samples, and blue indicates the group of type 2 diabetes blood samples.

Background: Type 2 diabetes (T2D) is a common chronic disease with many serious complications. Celastrol can prevent and treat type 2 diabetes by reversing insulin resistance in a number of ways. However, the specific mechanisms by which celastrol prevents and treats T2D are not well understood. The aim of this study was to explore the key gene targets and potential signaling pathway mechanisms of celastrol for the treatment of T2D.

Methods: GSE184050 was downloaded from the Gene Expression Omnibus online database. Blood samples from patients and healthy individuals with T2D were analyzed to identify differentially expressed genes (DEGs), and a protein−protein interaction network (PPI) was constructed. Key gene analysis of DEGs was performed using the MCODE plugin in Cystoscope as well as the Hubba plugin, and intersections were taken to obtain hub genes, which were displayed using a Venn diagram. Enrichment analysis was then performed via the ClueGo plugin in Cytoscape and validated using Gene Set Enrichment Analysis. The therapeutic targets of celastrol were then analyzed by pharmacophore network pharmacology, intersected to identify the therapeutic targets of celastrol, enriched for all targets, and intersected to obtain the signaling pathways for celastrol treatment. The protein structures of the therapeutic targets were predicted using the artificial intelligence AlphaFold2. Finally, molecular docking was used to verify whether celastrol could be successfully docked to the predicted targets.

Results: 618 DEGs were obtained, and 9 hub genes for T2D were identified by the MCODE and Hubba plug-ins, including ADAMTS15, ADAMTS7, ADAMTSL1, SEMA5B, ADAMTS8, THBS2, HBB, HBD and HBG2. The DEG-enriched signaling pathways mainly included the ferroptosis and TGF-beta signaling pathways. A total of 228 target genes were annotated by pharmacophore target analysis, and the therapeutic targets were identified, including S100A11, RBP3, HBB, BMP7 and IQUB, and 9 therapeutic signaling pathways were obtained by an intersectional set. The protein structures of the therapeutic targets were successfully predicted by AlphaFold2, and docking was validated using molecular docking.

Conclusion: Celastrol may prevent and treat T2D through key target genes, such as HBB, as well as signaling pathways, such as the TGF-beta signaling pathway and type II diabetes mellitus.

6,115 views

12 citations