Biology subjects

Chan, L.

Publications and source records attributed to Chan, L..

3 recordsLinked to original sources

bioRxiv2018

Unsupervised Machine learning to subtype Sepsis-Associated Acute Kidney Injury

ObjectiveAcute kidney injury (AKI) is highly prevalent in critically ill patients with sepsis. Sepsis-associated AKI is a heterogeneous clinical entity, and, like many complex syndromes, is composed of distinct subtypes. We aimed to agnostically identify AKI subphenotypes using machine learning techniques and routinely collected data in electronic health records (EHRs).\n\nDesignCohort study utilizing the MIMIC-III Database.\n\nSettingICUs from tertiary care hospital in the U.S.\n\nPatientsPatients older than 18 years with sepsis and who developed AKI within 48 hours of ICU admission.\n\nInterventionsUnsupervised machine learning utilizing all available vital signs and laboratory measurements.\n\nMeasurements and Main ResultsWe identified 1,865 patients with sepsis-associated AKI. Ten vital signs and 691 unique laboratory results were identified. After data processing and feature selection, 59 features, of which 28 were measures of intra-patient variability, remained for inclusion into an unsupervised machine-learning algorithm. We utilized k-means clustering with k ranging from 2 - 10; k=2 had the highest silhouette score (0.62). Cluster 1 had 1,358 patients while Cluster 2 had 507 patients. There were no significant differences between clusters on age, race or gender. We found significant differences in comorbidities and small but significant differences in several laboratory variables (hematocrit, bicarbonate, albumin) and vital signs (systolic blood pressure and heart rate). In-hospital mortality was higher in cluster 2 patients, 25% vs. 20%, p=0.008. Features with the largest differences between clusters included variability in basophil and eosinophil counts, alanine aminotransferase levels and creatine kinase values.\n\nConclusionsUtilizing routinely collected laboratory variables and vital signs in the EHR, we were able to identify two distinct subphenotypes of sepsis-associated AKI with different outcomes. Variability in laboratory variables, as opposed to their actual value, was more important for determination of subphenotypes. Our findings show the potential utility of unsupervised machine learning to better subtype AKI.

bioinformatics↗

bioRxiv2018

A cross-organism framework for supervised enhancer prediction with epigenetic pattern recognition and targeted validation

Enhancers are important noncoding elements, but they have been traditionally hard to characterize experimentally. Only a few mammalian enhancers have been validated, making it difficult to train statistical models for their identification properly. Instead, postulated patterns of genomic features have been used heuristically for identification. The development of massively parallel assays allows for the characterization of large numbers of enhancers for the first time. Here, we developed a framework that uses Drosophila STARR-seq data to create shape-matching filters based on enhancer-associated meta-profiles of epigenetic features. We combined these features with supervised machine learning algorithms (e.g., support vector machines) to predict enhancers. We demonstrated that our model could be applied to predict enhancers in mammalian species (i.e., mouse and human). We comprehensively validated the predictions using a combination of in vivo and in vitro approaches, involving transgenic assays in mouse and transduction-based reporter assays in human cell lines. Overall, the validations involved 153 enhancers in 6 mouse tissues and 4 human cell lines. The results confirmed that our model can accurately predict enhancers in different species without re-parameterization. Finally, we examined the transcription-factor binding patterns at predicted enhancers and promoters in human cell lines. We demonstrated that these patterns enable the construction of a secondary model effectively discriminating between enhancers and promoters.

bioinformatics↗

bioRxiv2018

Is it actionable? An Evaluation of the Rapid PCR-Based Blood Culture Identification Panel on the Management of Gram-Positive and Gram-Negative Blood Stream Infection

BackgroundThere is growing interest in the use of rapid blood culture identification (BCID) panels in antimicrobial stewardship programs (ASP). While many studies have looked at its clinical and economic utility, its comparative utility in gram-positive and gram-negative blood stream infections (BSI) have not been as well characterized.\n\nMethodsThe study was a quasi-experimental retrospective study at the Mayo Clinic in Phoenix, Arizona. All adult patients with positive blood cultures before BCID implementation (June 2015 to December 2015) and after BCID implementation (June 2016 to December 2016) were included. The outcomes of interest included: time to first appropriate antibiotic escalation, time to first appropriate antibiotic de-escalation, time to organism identification, LOS, infectious disease consultation, discharge disposition, and in-hospital mortality.\n\nResultsIn total, 203 patients were included in this study. There was a significant difference in the time to organism identification between pre- and post-BCID cohorts (27.1h vs. 3.3h, p<0.0001). BCID did not significantly reduce the time to first appropriate antimicrobial escalation or de-escalation for either GP-BSIs or GN-BSIs. Providers were more likely to escalate antimicrobial therapy in GP-BSIs after gram stain and more likely to de-escalate therapy in GN-BSIs after susceptibilities. While there were no significant differences in changes in antimicrobial therapy after organism identification by BCID, over a quarter of providers (28.1%) made changes after organism identification.\n\nConclusionsWhile BCID significantly reduced the time to identification for both GP-BSIs and GN-BSIs, BCID did not reduce the time to first appropriate antimicrobial escalation and de-escalation.

microbiology↗