Rectified Classifier Chains for Prediction of Antibiotic Resistance from Multi-labelled Data with Missing Labels

IEEE/ACM Transactions on Computational Biology and Bioinformatics, Journal Year: 2022, Volume and Issue: 20(1), P. 625 - 636

Published: Feb. 7, 2022

Predicting Antimicrobial Resistance (AMR) from genomic data has important implications for human and animal healthcare, especially given its potential more rapid diagnostics informed treatment choices. With the recent advances in sequencing technologies, applying machine learning techniques AMR prediction have indicated promising results. Despite this, there are shortcomings literature concerning methodologies suitable multi-drug where samples with missing labels exist. To address this shortcoming, we introduce a Rectified Classifier Chain (RCC) method predicting resistance. This RCC was tested using annotated features of genomics sequences compared similar multi-label classification methodologies. We found that eXtreme Gradient Boosting (XGBoost) base model to our outperformed second-best model, XGBoost based binary relevance by 3.3% Hamming accuracy 7.8% F1-score. Additionally, note models applied typically unsuitable identifying biomarkers informative their decisions; study, show contributing can also be identified proposed method. expect facilitate genome annotation pave path towards new indicative AMR.

Language: Английский

---

Predictive Modeling of Phenotypic Antimicrobial Susceptibility of Selected Beta-Lactam Antimicrobials from Beta-Lactamase Resistance Genes

Antibiotics, Journal Year: 2024, Volume and Issue: 13(3), P. 224 - 224

Published: Feb. 28, 2024

The outcome of bacterial infection management relies on prompt diagnosis and effective treatment, but conventional antimicrobial susceptibility testing can be slow labor-intensive. Therefore, this study aims to predict phenotypic selected beta-lactam antimicrobials in the bacteria family Enterobacteriaceae from different beta-lactamase resistance genotypes. Using human datasets extracted Antimicrobial Testing Leadership Surveillance (ATLAS) program conducted by Pfizer retail meat National Resistance Monitoring System for Enteric Bacteria (NARMS), we used a robust or weighted least square multivariable linear regression modeling framework explore relationship between data types genes. In humans, presence blaCTX-M-1, blaCTX-M-2, blaCTX-M-8/25, blaCTX-M-9 groups, MICs cephalosporins significantly increased values 0.34–3.07 μg/mL, however, carbapenem decreased 0.81–0.87 μg/mL. carbapenemase genes (blaKPC, blaNDM, blaIMP, blaVIM), cephalosporin 1.06–5.77 while 5.39–67.38 meat, MIC ceftriaxone blaCMY-2, blaCTX-M-55, blaCTX-M-65, blaSHV-2 55.16 222.70 250.81 204.89 31.51 μg/mL respectively. cefoxitin blaCTX-M-65 blaTEM-1 1.57 1.04 an average 8.66 over 17 years. Compared E. coli isolates, Salmonella enterica isolates 0.67 On other hand, ceftiofur blaSHV-2, 8.82 9.11 8.18 10.20 14 ability directly may help reduce reliance routine with higher turnaround times diagnostic, therapeutic, surveillance antimicrobial-resistant Enterobacteriaceae.

Language: Английский

---

Multi-excitation Raman Spectroscopy Complements Whole Genome Sequencing for Rapid Detection of Bacterial Infection and Resistance in WHO Priority Pathogens

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2022, Volume and Issue: unknown

Published: Feb. 8, 2022

Abstract Current methods for diagnosing acute and complex infections mostly rely on culture-based and, biofilms, fluorescence in-situ hybridization. These techniques are labor-intensive can take 2-4 days to return a test result, especially considering an extra culturing step required the antibiotic susceptibility testing (AST). This places significant burden healthcare providers, delaying treatment leading adverse patient outcomes. Here, we report complementary use of our newly developed multi-excitation Raman spectroscopy (ME-RS) method with whole-genome sequencing (WGS). Four WHO priority pathogens AST phenotyped their antimicrobial resistance (AMR) profile determined by WGS. On application ME-RS find high correlation WGS characterization. Highly accurate classification based species (98.93%), wild-type/non-wild type (99.45%), presence or absence thick peptidoglycan layers in cell walls (100%), as well at individual strain level (99.29%). results clearly demonstrate potential rapid first-stage tool species, strain-level which be followed up confirmation. Such workflow facilitate efficient stewardship handle prevent spread AMR.

Language: Английский

---

Highly sensitive quantitative phase microscopy and deep learning complement whole genome sequencing for rapid detection of infection and antimicrobial resistance

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2022, Volume and Issue: unknown

Published: July 7, 2022

Abstract The current state-of-the-art infection and antimicrobial resistance diagnostics (AMR) is based mainly on culture-based methods with a detection time of 48-96 hours. Slow diagnoses lead to adverse patient outcomes that directly correlate the taken administer optimal antimicrobials. Mortality risk doubles 24-hour delay in providing appropriate antibiotics cases bacteremia. Therefore, it essential develop novel can promptly accurately diagnose microbial infections at both species strain levels clinical settings. Here, we demonstrate complimentary use label-free optical assay whole-genome sequencing (WGS) enable high-speed culture-free diagnosis AMR. Our microscopy exploiting label-free, highly sensitive quantitative phase (QPM) followed by deep convolutional neural networks (DCNNs) classification. We benchmarked our proposed workflow 21 isolates from four WHO priority pathogens ( Escherichia coli, Staphylococcus aureus, Klebsiella pneumoniae , Acinetobacter baumannii ) were antibiotic susceptibility testing (AST) phenotyped, their profile was determined WGS. good agreement WGS characterization. Highly accurate classification gram staining (100% for gram-negative 83.4% gram-positive), (98.6%), resistant/susceptible type (96.4%), as well individual level predicting 19 out strains). These results potential QPM rapid first-stage tool species, presence, absence AMR, strain-level classification, which follow up confirmation pathogen ID characterization AMR antibiotic. Taken together, all this information high importance. Such could potentially facilitate efficient stewardship prevent spread

Language: Английский

---

Explainable Deep Learning Approach for Multilabel Classification of Antimicrobial Resistance With Missing Labels

IEEE Access, Journal Year: 2022, Volume and Issue: 10, P. 113073 - 113085

Published: Jan. 1, 2022

Predicting Antimicrobial Resistance (AMR) from genomic sequence data has become a significant component of overcoming the AMR challenge, especially given its potential for facilitating more rapid diagnostics and personalised antibiotic treatments. With recent advances in sequencing technologies computing power, deep learning models have been widely adopted to predict reliably error-free. There are many different types AMR; therefore, any practical prediction system must be able identify multiple AMRs present sequence. Unfortunately, most datasets do not all labels marked, thereby making modelling approach challenging owing reliance on reliability accuracy. This paper addresses this issue by presenting an effective solution, Mask-Loss 1D convolution neural network (ML-ConvNet), with missing labels. The core ML- ConvNet utilises masked loss function that overcomes effect predicting AMR. proposed ML-ConvNet is demonstrated outperform state-of-the-art methods literature 10.5%, according F1 score. model's performance evaluated using degrees label found conventional 76% score when 86.68% missing. Furthermore, was established explainable artificial intelligence (XAI) pipeline, it ideally suited hospital healthcare settings, where model interpretability essential requirement.

Language: Английский

---