MSDeepAMR: antimicrobial resistance prediction based on deep neural networks and transfer learning
IntroductionAntimicrobial resistance (AMR) is a global health problem that requires early and effective treatments to prevent the indiscriminate use of antimicrobial drugs and the outcome of infections. Mass Spectrometry (MS), and more particularly MALDI-TOF, have been widely adopted by routine clin...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2024-04-01
|
| Series: | Frontiers in Microbiology |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fmicb.2024.1361795/full |
| _version_ | 1797202436900782080 |
|---|---|
| author | Xaviera A. López-Cortés Xaviera A. López-Cortés José M. Manríquez-Troncoso Ruber Hernández-García Ruber Hernández-García Daniel Peralta |
| author_facet | Xaviera A. López-Cortés Xaviera A. López-Cortés José M. Manríquez-Troncoso Ruber Hernández-García Ruber Hernández-García Daniel Peralta |
| author_sort | Xaviera A. López-Cortés |
| collection | DOAJ |
| description | IntroductionAntimicrobial resistance (AMR) is a global health problem that requires early and effective treatments to prevent the indiscriminate use of antimicrobial drugs and the outcome of infections. Mass Spectrometry (MS), and more particularly MALDI-TOF, have been widely adopted by routine clinical microbiology laboratories to identify bacterial species and detect AMR. The analysis of AMR with deep learning is still recent, and most models depend on filters and preprocessing techniques manually applied on spectra.MethodsThis study propose a deep neural network, MSDeepAMR, to learn from raw mass spectra to predict AMR. MSDeepAMR model was implemented for Escherichia coli, Klebsiella pneumoniae, and Staphylococcus aureus under different antibiotic resistance profiles. Additionally, a transfer learning test was performed to study the benefits of adapting the previously trained models to external data.ResultsMSDeepAMR models showed a good classification performance to detect antibiotic resistance. The AUROC of the model was above 0.83 in most cases studied, improving the results of previous investigations by over 10%. The adapted models improved the AUROC by up to 20% when compared to a model trained only with external data.DiscussionThis study demonstrate the potential of the MSDeepAMR model to predict antibiotic resistance and their use on external MS data. This allow the extrapolation of the MSDeepAMR model to de used in different laboratories that need to study AMR and do not have the capacity for an extensive sample collection. |
| first_indexed | 2024-04-24T08:03:25Z |
| format | Article |
| id | doaj.art-fdbeb89e1e8e44de8293fe6ef3248a25 |
| institution | Directory Open Access Journal |
| issn | 1664-302X |
| language | English |
| last_indexed | 2024-04-24T08:03:25Z |
| publishDate | 2024-04-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Microbiology |
| spelling | doaj.art-fdbeb89e1e8e44de8293fe6ef3248a252024-04-17T15:42:20ZengFrontiers Media S.A.Frontiers in Microbiology1664-302X2024-04-011510.3389/fmicb.2024.13617951361795MSDeepAMR: antimicrobial resistance prediction based on deep neural networks and transfer learningXaviera A. López-Cortés0Xaviera A. López-Cortés1José M. Manríquez-Troncoso2Ruber Hernández-García3Ruber Hernández-García4Daniel Peralta5Department of Computer Sciences and Industries, Universidad Católica del Maule, Talca, ChileCentro de Innovación en Ingeniería Aplicada (CIIA), Universidad Católica del Maule, Talca, ChileDepartment of Computer Sciences and Industries, Universidad Católica del Maule, Talca, ChileDepartment of Computer Sciences and Industries, Universidad Católica del Maule, Talca, ChileLaboratory of Technological Research in Pattern Recognition (LITRP), Universidad Católica del Maule, Talca, ChileIDLab, Department of Information Technology, Ghent University-imec, Ghent, BelgiumIntroductionAntimicrobial resistance (AMR) is a global health problem that requires early and effective treatments to prevent the indiscriminate use of antimicrobial drugs and the outcome of infections. Mass Spectrometry (MS), and more particularly MALDI-TOF, have been widely adopted by routine clinical microbiology laboratories to identify bacterial species and detect AMR. The analysis of AMR with deep learning is still recent, and most models depend on filters and preprocessing techniques manually applied on spectra.MethodsThis study propose a deep neural network, MSDeepAMR, to learn from raw mass spectra to predict AMR. MSDeepAMR model was implemented for Escherichia coli, Klebsiella pneumoniae, and Staphylococcus aureus under different antibiotic resistance profiles. Additionally, a transfer learning test was performed to study the benefits of adapting the previously trained models to external data.ResultsMSDeepAMR models showed a good classification performance to detect antibiotic resistance. The AUROC of the model was above 0.83 in most cases studied, improving the results of previous investigations by over 10%. The adapted models improved the AUROC by up to 20% when compared to a model trained only with external data.DiscussionThis study demonstrate the potential of the MSDeepAMR model to predict antibiotic resistance and their use on external MS data. This allow the extrapolation of the MSDeepAMR model to de used in different laboratories that need to study AMR and do not have the capacity for an extensive sample collection.https://www.frontiersin.org/articles/10.3389/fmicb.2024.1361795/fullMALDI-TOFdeep learningantibiotic resistanceEscherichia coliKlebsiella pneumoniaeStaphylococcus aureus |
| spellingShingle | Xaviera A. López-Cortés Xaviera A. López-Cortés José M. Manríquez-Troncoso Ruber Hernández-García Ruber Hernández-García Daniel Peralta MSDeepAMR: antimicrobial resistance prediction based on deep neural networks and transfer learning Frontiers in Microbiology MALDI-TOF deep learning antibiotic resistance Escherichia coli Klebsiella pneumoniae Staphylococcus aureus |
| title | MSDeepAMR: antimicrobial resistance prediction based on deep neural networks and transfer learning |
| title_full | MSDeepAMR: antimicrobial resistance prediction based on deep neural networks and transfer learning |
| title_fullStr | MSDeepAMR: antimicrobial resistance prediction based on deep neural networks and transfer learning |
| title_full_unstemmed | MSDeepAMR: antimicrobial resistance prediction based on deep neural networks and transfer learning |
| title_short | MSDeepAMR: antimicrobial resistance prediction based on deep neural networks and transfer learning |
| title_sort | msdeepamr antimicrobial resistance prediction based on deep neural networks and transfer learning |
| topic | MALDI-TOF deep learning antibiotic resistance Escherichia coli Klebsiella pneumoniae Staphylococcus aureus |
| url | https://www.frontiersin.org/articles/10.3389/fmicb.2024.1361795/full |
| work_keys_str_mv | AT xavieraalopezcortes msdeepamrantimicrobialresistancepredictionbasedondeepneuralnetworksandtransferlearning AT xavieraalopezcortes msdeepamrantimicrobialresistancepredictionbasedondeepneuralnetworksandtransferlearning AT josemmanriqueztroncoso msdeepamrantimicrobialresistancepredictionbasedondeepneuralnetworksandtransferlearning AT ruberhernandezgarcia msdeepamrantimicrobialresistancepredictionbasedondeepneuralnetworksandtransferlearning AT ruberhernandezgarcia msdeepamrantimicrobialresistancepredictionbasedondeepneuralnetworksandtransferlearning AT danielperalta msdeepamrantimicrobialresistancepredictionbasedondeepneuralnetworksandtransferlearning |