Artificial Neural Network Model for Liver Cirrhosis Diagnosis in Patients with Hepatitis B Virus-Related Hepatocellular Carcinoma

Machine Learning for Short-Term Mortality in Acute Decompensation of Liver Cirrhosis: Better than MELD Score

Prediction of short-term mortality in patients with acute decompensation of liver cirrhosis could be improved. We aimed to develop and validate two machine learning (ML) models for predicting 28-day and 90-day mortality in patients hospitalized with acute decompensated liver cirrhosis. We trained two artificial neural network (ANN)-based ML models using a training sample of 165 out of 290 (56.9%) patients, and then tested their predictive performance against Model of End-stage Liver Disease-Sodium (MELD-Na) and MELD 3.0 scores using a different validation sample of 125 out of 290 (43.1%) patients. The area under the ROC curve (AUC) for predicting 28-day mortality for the ML model was 0.811 (95%CI: 0.714- 0.907; p < 0.001), while the AUC for the MELD-Na score was 0.577 (95%CI: 0.435-0.720; p = 0.226) and for MELD 3.0 was 0.600 (95%CI: 0.462-0.739; p = 0.117). The area under the ROC curve (AUC) for predicting 90-day mortality for the ML model was 0.839 (95%CI: 0.776- 0.884; p < 0.001), while the AUC for the MELD-Na score was 0.682 (95%CI: 0.575-0.790; p = 0.002) and for MELD 3.0 was 0.703 (95%CI: 0.590-0.816; p < 0.001). Our study demonstrates that ML-based models for predicting short-term mortality in patients with acute decompensation of liver cirrhosis perform significantly better than MELD-Na and MELD 3.0 scores in a validation cohort.

Machine learning-based model for predicting the esophagogastric variceal bleeding risk in liver cirrhosis patients

BACKGROUND

Liver cirrhosis patients are at risk for esophagogastric variceal bleeding (EGVB). Herein, we aimed to estimate the EGVB risk in patients with liver cirrhosis using an artificial neural network (ANN).

METHODS

We included 999 liver cirrhosis patients hospitalized at the Beijing Ditan Hospital, Capital Medical University in the training cohort and 101 patients from Shuguang Hospital in the validation cohort. The factors independently affecting EGVB occurrence were determined via univariate analysis and used to develop an ANN model.

RESULTS

The 1-year cumulative EGVB incidence rates were 11.9 and 11.9% in the training and validation groups, respectively. A total of 12 independent risk factors, including gender, drinking and smoking history, decompensation, ascites, location and size of varices, alanine aminotransferase (ALT), γ-glutamyl transferase (GGT), hematocrit (HCT) and neutrophil-lymphocyte ratio (NLR) levels as well as red blood cell (RBC) count were evaluated and used to establish the ANN model, which estimated the 1-year EGVB risk. The ANN model had an area under the curve (AUC) of 0.959, which was significantly higher than the AUC for the North Italian Endoscopic Club (NIEC) (0.669) and revised North Italian Endoscopic Club (Rev-NIEC) indices (0.725) (all P <  0.001). Decision curve analyses revealed improved net benefits of the ANN compared to the NIEC and Rev-NIEC indices.

CONCLUSIONS

The ANN model accurately predicted the 1-year risk for EGVB in liver cirrhosis patients and might be used as a basis for risk-based EGVB surveillance strategies.

Artificial intelligence model with deep learning in nonalcoholic fatty liver disease diagnosis: genetic based artificial neural networks

Nonalcoholic fatty liver disease (NAFLD) is one of the most common causes of chronic liver disease in the world. The NAFLD spectrum includes simple steatosis, steatohepatitis, fibrosis, cirrhosis, and hepatocellular carcinoma (HCC). Genetic, nutritional factors, obesity, insulin resistance, gut microbiota are among the risk factors for NAFLD. The genetic variant Patatin-like phospholipase domain-containing protein 3 (PNPLA3) plays an important role in the development of a number of liver diseases ranging from steatosis, chronic hepatitis, cirrhosis and HCC. Due to the increase in the prevalence of NAFLD, new models are being developed with machine learning, deep learning, artificial neural network (ANN) algorithms in the field of artificial intelligence (AI) to determine low-cost, noninvasive diagnostic methods. Models developed with ANN from AI modules are important in order to examine biochemical and genomic information in detail in the diagnosis of NAFLD. The aim of this study is to develop a simple ANN model using biochemical and genotypic parameters in the diagnosis of NAFLD. A total of 300 patients followed up with the diagnosis of NAFLD and 100 controls were included in the study. The data set was divided into two as training and test set. Genotyping of PNPLA3 (CC, CG, GG) as genomic analysis was performed with real time PCR device. The algorithm used for the diagnosis of NAFLD was designed using age, body mass index (BMI), mean platelet volume (MPV), insulin resistance (IR), alanine aminotransferase (ALT), genotype PNPLA3 (CC, CG, GG) parameters. MLP Classifier algorithm from ANN was used in the development of the model. ANN algorithms are used in python programming language. Statistical analyzes were made in SPSS program. Percent accuracy, area under the ROC curve, confusion matrix, Positive (PPV) and Negative Predicted Value (NPV) values, precision, recall, and f1-score results were determined. The accuracy percentage was determined as 0.979 in the train set and 0.970 in the test set. The Log Loss value was set to 0.09. The developed neural network achieved an accuracy percentage of 97.0% during testing, with an area under the ROC curve value of 0.95. We think that the ANN model developed with genomic and biochemical parameters can be used as a cost-effective, noninvasive new predictive diagnostic model in clinical practice in the diagnosis of NAFLD.

A novel diagnostic four-gene signature for hepatocellular carcinoma based on artificial neural network: Development, validation, and drug screening

Background: Hepatocellular carcinoma (HCC) is one of the most common cancers with high mortality in the world. HCC screening and diagnostic models are becoming effective strategies to reduce mortality and improve the overall survival (OS) of patients. Here, we expected to establish an effective novel diagnostic model based on new genes and explore potential drugs for HCC therapy.

Methods: The gene expression data of HCC and normal samples (GSE14811, GSE60502, GSE84402, GSE101685, GSE102079, GSE113996, and GSE45436) were downloaded from the Gene Expression Omnibus (GEO) dataset. Bioinformatics analysis was performed to distinguish two differentially expressed genes (DEGs), diagnostic candidate genes, and functional enrichment pathways. QRT-PCR was used to validate the expression of diagnostic candidate genes. A diagnostic model based on candidate genes was established by an artificial neural network (ANN). Drug sensitivity analysis was used to explore potential drugs for HCC. CCK-8 assay was used to detect the viability of HepG2 under various presentative chemotherapy drugs.

Results: There were 82 DEGs in cancer tissues compared to normal tissue. Protein–protein interaction (PPI), Gene Ontology (GO), and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses and infiltrating immune cell analysis were administered and analyzed. Diagnostic-related genes of MT1M, SPINK1, AKR1B10, and SLCO1B3 were selected from DEGs and used to construct a diagnostic model. The receiver operating characteristic (ROC) curves were 0.910 and 0.953 in the training and testing cohorts, respectively. Potential drugs, including vemurafenib, LOXO-101, dabrafenib, selumetinib, Arry-162, and NMS-E628, were found as well. Vemurafenib, dabrafenib, and selumetinib were observed to significantly affect HepG2 cell viability.

Conclusion: The diagnostic model based on the four diagnostic-related genes by the ANN could provide predictive significance for diagnosis of HCC patients, which would be worthy of clinical application. Also, potential chemotherapy drugs might be effective for HCC therapy.

Effect of mutations across reverse transcriptase region on HBV replication and progression of liver diseases in Chinese patients

It was known that mutations in the RT region were mainly related to nucleot(s)ide analogs resistance. Increasing studies indicated that RT mutations were related to advanced liver diseases (ALD) and had effects on HBV replication, but the distribution characteristics of mutations across RT region in the development of liver diseases and the effect of RT mutations on HBV replication were not fully clarified. HBV RT region was direct-sequenced in 1473 chronic HBV-infected patients. Mutation frequencies were analyzed to identify the specific mutations differing between groups classified by genotypes, loads of HBV DNA, or progression of liver diseases. In the range of rt145-rt290, rt145, rt221, rt222, rt267, and rt271 were the genotype-polymorphic sites, while rt238 was the genotype-specific sites. Mutations at rt163, rt173, rt180, rt181, rt184, rt191, rt199, and rt214 were more frequent among patients with C-genotype HBV, while those at rt220, rt225, rt226, rt269, and rt274 were more frequent among patients with B-genotype HBV. RtM204V/I could reduce the HBV DNA loads while rtQ/L267H/R could increase the HBV DNA loads. RtV214A/E/I (OR 3.94, 95% CI 1.09 to 14.26) was an independent risk factor for advanced liver diseases. In summary, the hotspots of mutations were different between B and C genotypes. Besides the effect on the S region, RT mutations had effects on HBV replication by other unknown ways. RtV214A/E/I was found to be an independent risk factor for ALD, suggesting that mutations at rt214 site could be used as a potential virological marker for the liver disease progression.

Artificial neural network model to predict post-hepatectomy early recurrence of hepatocellular carcinoma without macroscopic vascular invasion

BACKGROUND

The accurate prediction of post-hepatectomy early recurrence (PHER) of hepatocellular carcinoma (HCC) is vital in determining postoperative adjuvant treatment and monitoring. This study aimed to develop and validate an artificial neural network (ANN) model to predict PHER in HCC patients without macroscopic vascular invasion.

METHODS

Nine hundred and three patients who underwent curative liver resection for HCC participated in this study. They were randomly divided into derivation (n = 679) and validation (n = 224) cohorts. The ANN model was developed in the derivation cohort and subsequently verified in the validation cohort.

RESULTS

PHER morbidity in the derivation and validation cohorts was 34.8 and 39.2%, respectively. A multivariable analysis revealed that hepatitis B virus deoxyribonucleic acid load, γ-glutamyl transpeptidase level, α-fetoprotein level, tumor size, tumor differentiation, microvascular invasion, satellite nodules, and blood loss were significantly associated with PHER. These factors were incorporated into an ANN model, which displayed greater discriminatory abilities than a Cox's proportional hazards model, preexisting recurrence models, and commonly used staging systems for predicting PHER. The recurrence-free survival curves were significantly different between patients that had been stratified into two risk groups.

CONCLUSION

When compared to other models and staging systems, the ANN model has a significant advantage in predicting PHER for HCC patients without macroscopic vascular invasion.