Dataset Information

An integrated pipeline for prediction of Clostridioides difficile infection.

ABSTRACT: With the expansion of electronic health records(EHR)-linked genomic data comes the development of machine learning-enable models. There is a pressing need to develop robust pipelines to evaluate the performance of integrated models and minimize systemic bias. We developed a prediction model of symptomatic Clostridioides difficile infection(CDI) by integrating common EHR-based and genetic risk factors(rs2227306/IL8). Our pipeline includes (1) leveraging phenotyping algorithm to minimize temporal bias, (2) performing simulation studies to determine the predictive power in samples without genetic information, (3) propensity score matching to control for the confoundings, (4) selecting machine learning algorithms to capture complex feature interactions, (5) performing oversampling to address data imbalance, and (6) optimizing models and ensuring proper bias-variance trade-off. We evaluate the performance of prediction models of CDI when including common clinical risk factors and the benefit of incorporating genetic feature(s) into the models. We emphasize the importance of building a robust integrated pipeline to avoid systemic bias and thoroughly evaluating genetic features when integrated into the prediction models in the general population and subgroups.

SUBMITTER: Li J

PROVIDER: S-EPMC10545794 | biostudies-literature | 2023 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

An integrated pipeline for prediction of Clostridioides difficile infection.

Li Jiang J Chaudhary Durgesh D Sharma Vaibhav V Sharma Vishakha V Avula Venkatesh V Ssentongo Paddy P Wolk Donna M DM Zand Ramin R Abedi Vida V

Scientific reports 20231002 1

With the expansion of electronic health records(EHR)-linked genomic data comes the development of machine learning-enable models. There is a pressing need to develop robust pipelines to evaluate the performance of integrated models and minimize systemic bias. We developed a prediction model of symptomatic Clostridioides difficile infection(CDI) by integrating common EHR-based and genetic risk factors(rs2227306/IL8). Our pipeline includes (1) leveraging phenotyping algorithm to minimize temporal ...[more]

PMID: 37783691

Similar Datasets

Project description:There has been no prior application of matched metagenomics and metatranscriptomics in Clostridioides difficile infection (CDI) evaluating the role of fungi in CDI or identifying community functions that contribute to the development of this disease. We collected diarrheal stools from 49 inpatients (18 of whom tested positive for CDI) under stringent inclusion criteria. We utilized a tiered sequencing approach to identify enriched bacterial and fungal taxa, using 16S and internal transcribed spacer (ITS) rRNA gene amplicon sequencing, with matched metagenomics and metatranscriptomics performed on a subset of the population. Distinct bacterial and fungal compositions distinguished CDI-positive and -negative patients, with the greatest differentiation between the cohorts observed based on bacterial metatranscriptomics. Bipartite network analyses demonstrated that Aspergillus and Penicillium taxa shared a strong positive relationship in CDI patients and together formed negative cooccurring relationships with several bacterial taxa, including the Oscillospira, Comamonadaceae, Microbacteriaceae, and Cytophagaceae Metatranscriptomics revealed enriched pathways in CDI patients associated with biofilm production primarily driven by Escherichia coli and Pseudomonas, quorum-sensing proteins, and two-component systems related to functions such as osmotic regulation, linoleic acid metabolism, and flagellar assembly. Differential expression of functional pathways unveiled a mechanism by which the causal dysbiosis of CDI may self-perpetuate, potentially contributing to treatment failures. We propose that CDI has a distinct fungus-associated bacteriome, and this first description of metatranscriptomics in human subjects with CDI demonstrates that inflammation, osmotic changes, and biofilm production are key elements of CDI pathophysiology.IMPORTANCE Our data suggest a potential role for fungi in the most common nosocomial bacterial infection in the United States, introducing the concept of a transkingdom interaction between bacteria and fungi in this disease. We also provide the first direct measure of microbial community function in Clostridioides difficile infection using patient-derived tissue samples, revealing antibiotic-independent mechanisms by which C. difficile infection may resist a return to a healthy gut microbiome.

Dataset Information

An integrated pipeline for prediction of Clostridioides difficile infection.

Publications

An integrated pipeline for prediction of Clostridioides difficile infection.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets