Early prediction of sepsis-induced respiratory tract infection using a biomarker-based machine-learning algorithm.

Early and differential diagnosis of sepsis is essential to avoid unnecessary antibiotic use and further reduce patient morbidity and mortality. Here, we aimed to identify predictors of sepsis and advance a machine-learning strategy to predict sepsis-induced respiratory tract infection (RTI). Patients with sepsis and RTI were selected via retrospective analysis, and essential population characteristics and laboratory parameters were recorded. To improve the performance of the primary model and avoid over-fitting, a recursive feature elimination with cross-validation (RFECV) strategy was used to screen the optimal subset of biomarkers and construct nine machine-learning models based on this subset; the average accuracy, precision, recall, and F1-score were used for evaluation of the models. We identified 430 patients with sepsis and 686 patients with RTI. A total of 39 features were collected, with 23 features identified for initial model construction. Using the RFECV algorithm, we found that the XGBoost classifier, which only needed to include seven biomarkers, demonstrated the best performance among all prediction models, with an average accuracy of 89.24 ± 2.28, while the Ridge classifier, which included 11 biomarkers, had an average accuracy of only 83.87 ± 4.69. The remaining models had prediction accuracies greater than 88%. We developed nine models for predicting sepsis using a strategy that combined RFECV with machine learning. Among these models, the XGBoost classifier, which included seven biomarkers, showed the best performance and highest accuracy for predicting sepsis and may be a promising tool for the timely identification of sepsis.

Developing a machine learning prediction algorithm for early differentiation of urosepsis from urinary tract infection.

OBJECTIVES: Early recognition and timely intervention for urosepsis are key to reducing morbidity and mortality. Blood culture has low sensitivity, and a long turnaround time makes meeting the needs of clinical diagnosis difficult. This study aimed to use biomarkers to build a machine learning model for early prediction of urosepsis. METHODS: Through retrospective analysis, we screened 157 patients with urosepsis and 417 patients with urinary tract infection. Laboratory data of the study participants were collected, including data on biomarkers, such as procalcitonin, D-dimer, and C-reactive protein. We split the data into training (80%) and validation datasets (20%) and determined the average model prediction accuracy through cross-validation. RESULTS: In total, 26 variables were initially screened and 18 were statistically significant. The influence of the 18 variables was sorted using three ranking methods to further determine the best combination of variables. The Gini importance ranking method was found to be suitable for variable filtering. The accuracy rates of the six machine learning models in predicting urosepsis were all higher than 80%, and the performance of the artificial neural network (ANN) was the best among all. When the ANN included the eight biomarkers with the highest influence ranking, its model had the best prediction performance, with an accuracy rate of 92.9% and an area under the receiver operating characteristic curve of 0.946. CONCLUSIONS: Urosepsis can be predicted using only the top eight biomarkers determined by the ranking method. This data-driven predictive model will enable clinicians to make quick and accurate diagnoses.

Genetic variants in class II transactivator are associated with chronic hepatitis B virus infection in the Han Chinese population.

Class II transactivator (CIITA) is a master regulator of MHC gene expression and plays a role in inducing the expression of other immune system genes, including IL-4, IL-10 and Fas ligand, as well as more than 60 other immunologically significant genes. We used CIITA as a candidate gene to analyse whether any single-nucleotide polymorphisms (SNPs) are associated with chronic hepatitis B virus (HBV) infection. In total, 773 patients with chronic HBV infection were enrolled in this hospital-based case-control study. The patients were divided into groups according to their clinical characteristics: 596 patients had chronic hepatitis B (CHB), and 177 patients had hepatocellular carcinoma (HCC). A total of 313 patients with self-limited HBV infection were selected as the control group. CIITA gene variants were screened using Haploview 4.2 software; improved multiplex ligation detection reaction technology was then used for genotype detection, and HaploReg v4.1 was employed to predict the functions of 15 variants. The results showed that SNPs in introns in the CIITA gene, namely, rs13333382 (TT + TA vs. AA: p = .003, odds ratio (OR) = 0.65, 95% confidence interval (CI) = 0.49-0.87) and rs4780335 (CC + CG vs. GG: p = 9.40 × 10-5 , OR = 0.55, 95% CI = 0.41-0.74), were positively associated with self-limited HBV infection in the dominant genetic model. Additionally, SNP rs1139564 (TT + TC vs. CC: p = .002, OR = 1.61, 95% CI = 1.19-2.16) in the 3' untranslated region may increase the risk of CHB. According to in silico analysis, all three statistically significant variants act as transcription factor binding motifs. However, we did not find that these 15 mutations are associated with HCC risk. Therefore, we believe that CIITA is a susceptibility gene for CHB rather than for HCC.

[Studies on the association of single nucleotide polymorphisms of HLA-DP and DQ genes with the outcome of chronic hepatitis B virus infection].

OBJECTIVE: To investigate the association of single nucleotide polymorphisms in the HLA-DP and DQ genes with the outcome of chronic hepatitis B virus infection. METHODS: Two hundred and four healthy subjects, 255 clearance subjects, 204 asymptomatic HBV carriers (AsC), 136 chronic hepatitis B (CHB), 68 liver cirrhosis (LC) and hepatocellular carcinoma (HCC) were enrolled. Genotypes of rs3077, rs9277535 and rs2647050 were determined by sequence specific primers-PCR (PCR-SSP). RESULTS: By using healthy subjects and clearance subjects as the control groups, rs3077 and rs9277535 were significantly associated with chronic HBV infection under additive and dominant models (P< 0.05). Meanwhile, haplotypes GGA, AGA, AAA appeared to be protective factors against chronic HBV infection (P < 0.05). By using AsC as the control group, comparison with the CHB, LC and HCC groups showed no association of the 3 SNPs or haplotypes with the clinical outcome (P > 0.05). CONCLUSION: HLA-DP gene polymorphisms are strongly associated with chronic HBV infection. The presence of A allele at rs3077 and rs9277535 of the HLA-DP gene may decreased the risk for chronic HBV infection.

Four Biomarkers-Based Artificial Neural Network Model for Accurate Early Prediction of Bacteremia with Low-level Procalcitonin.

OBJECTIVE: Procalcitonin levels above 2.0 ng/mL are associated with a higher risk of severe sepsis. Bacteremia with procalcitonin levels lower than 2.0 ng/mL has not received much attention, and relevant prediction models are lacking. Herein, a panel of biomarkers was used to establish an early predictive model for bacteremia using machine learning approaches. METHODS: Retrospective evaluation of 487 non-bacteremia controls and 444 bacteremia patients with low-level procalcitonin was performed. Clinical data, including procalcitonin, interleukin-6, C-reactive protein, D-dimer, and lactic acid levels, as well as leukocyte, neutrophil, and platelet counts, were used to identify a panel of relevant variables to build a machine learning model. RESULTS: By comparing six prediction models, the performance of an artificial neural network (ANN) was found to be superior to that of other designed models, with a sensitivity of 0.82, a specificity of 0.85, and an accuracy rate of 83.5%. Furthermore, interleukin-6, procalcitonin, D-dimer, and lactic acid were found to be the most influential biomarkers with the potential to predict bacteremia. CONCLUSION: The ANN model described herein holds outstanding predictive performance, with the potential to provide real-time, data-driven predictions of bacteremia.

Meta-analysis of the correlation between the rs17401966 polymorphism in kinesin family member 1B and susceptibility to hepatitis B virus related hepatocellular carcinoma.

BACKGROUND/AIMS: The association between the kinesin family member 1B (KIF1B) gene polymorphism and the risk of hepatitis B virus-related hepatocellular carcinoma (HCC) has been investigated in many peer-reviewed studies. However, scholars have failed to replicate these results in validation tests. The purpose of the present study was to explore whether the KIF1B rs17401966 polymorphism was associated with susceptibility to HCC. METHODS: The results of case-controlled studies on the correlation between the KIF1B rs17401966 polymorphism and HCC susceptibility were collected using Google Scholar and the EMBASE, PubMed and CNKI databases. Based on inclusion and exclusion criteria, 5 papers with a total of 12 cohorts were included in this study. RESULTS: The 12 cohorts were integrated, and the results showed that the rs17401966 polymorphism reduced the risk for HCC under the allele, heterozygous, homozygous, and dominant models but not under the additive or recessive models. Moreover, the merged results showed strong heterogeneity, and the cumulative meta-analysis results were unreliable. A genetic differentiation analysis of the 12 cohorts found different degrees of genetic differentiation between the 5 cohorts in Zhang et al.'s study and the cohorts in the other studies. We further divided the 12 study cohorts into 2 subgroups based on fixation index value; however, the results of that analysis were inconsistent. CONCLUSIONS: The results of this meta-analysis were not able to verify the association between the KIF1B rs1740199 polymorphism and HCC risk. Therefore, a well-designed, large-scale, multicenter validation study is needed to confirm the relationship.

Clinical significance of periodic detection of hepatitis B virus YVDD mutation by ultrasensitive real-time amplification refractory mutation system quantitative PCR during lamivudine treatment in patients with chronic hepatitis B.

Monitoring hepatitis B virus (HBV) mutants periodically during nucleoside analogue treatment is of great clinical significance, particularly in persistently HBV DNA-positive patients. However, few studies have investigated the dynamic changes of HBV YMDD (Tyr-Met-Asp-Asp) and YVDD (Tyr-Val-Asp-Asp) populations in chronic hepatitis B (CHB) patients whilst undergoing lamivudine (LMV) treatment. In this study, we sought to investigate the dynamic changes of HBV YMDD and YVDD variants by ultrasensitive real-time amplification refractory mutation system quantitative PCR (RT-ARMS-qPCR) and evaluate its significance for changes in the treatment of CHB patients. RT-ARMS-qPCR was established and evaluated with standard recombinant plasmids. Fifteen CHB patients receiving LMV (100 mg daily) were consecutively recruited and followed up for 60 weeks. Serum samples were obtained from each patient at baseline and every 12 weeks. The total HBV DNA, HBV YMDD DNA and YVDD DNA levels were measured using RT-ARMS-qPCR at all given time points after treatment. Routine liver biochemistry parameters, including aspartate aminotransferase and alanine aminotransferase, were also measured every 12 weeks. The linear range of the assay was between 1×10(12) and 1×10(5) copies ml(-1). The low detection limit was 1×10(4) copies ml(-1). After 60 weeks of LMV treatment, nine patients experienced virological breakthrough. The YVDD variant could be detected 12-48 weeks before virological breakthrough. The YVDD variant was detected as the predominant population (range 69.4-100 %) in patients by the time virological breakthrough appeared. We concluded that RT-ARMS-qPCR was sensitive for the detection and quantification of low levels of HBV mutation. Periodic detection of HBV YM(V)DD every 12 weeks during LMV treatment is helpful for therapeutic decision making.

Establishment of real time allele specific locked nucleic acid quantitative PCR for detection of HBV YIDD (ATT) mutation and evaluation of its application.

BACKGROUND: Long-term use of nucleos(t)ide analogues can increase risk of HBV drug-resistance mutations. The rtM204I (ATT coding for isoleucine) is one of the most important resistance mutation sites. Establishing a simple, rapid, reliable and highly sensitive assay to detect the resistant mutants as early as possible is of great clinical significance. METHODS: Recombinant plasmids for HBV YMDD (tyrosine-methionine-aspartate-aspartate) and YIDD (tyrosine-isoleucine-aspartate-aspartate) were constructed by TA cloning. Real time allele specific locked nucleic acid quantitative PCR (RT-AS-LNA-qPCR) with SYBR Green I was established by LNA-modified primers and evaluated with standard recombinant plasmids, clinical templates (the clinical wild type and mutant HBV DNA mixture) and 102 serum samples from nucleos(t)ide analogues-experienced patients. The serum samples from a chronic hepatitis B (CHB) patient firstly received LMV mono therapy and then switched to LMV + ADV combined therapy were also dynamically analyzed for 10 times. RESULTS: The linear range of the assay was between 1×10(9) copies/µl and 1 × 10(2) copies/µl. The low detection limit was 1 × 10(1) copies/µl. Sensitivity of the assay were 10(-6), 10(-4) and 10(-2) in the wild-type background of 1 × 10(9) copies/µl, 1 × 10(7) copies/µl and 1 × 10(5) copies/µl, respectively. The sensitivity of the assay in detection of clinical samples was 0.03%. The complete coincidence rate between RT-AS-LNA-qPCR and direct sequencing was 91.2% (93/102), partial coincidence rate was 8.8% (9/102), and no complete discordance was observed. The two assays showed a high concordance (Kappa = 0.676, P = 0.000). Minor variants can be detected 18 weeks earlier than the rebound of HBV DNA load and alanine aminotransferase level. CONCLUSIONS: A rapid, cost-effective, high sensitive, specific and reliable method of RT-AS-LNA-qPCR with SYBR Green I for early and absolute quantification of HBV YIDD (ATT coding for isoleucine) variants was established, which can provide valuable information for clinical antiretroviral regimens.