Matrix-assisted laser desorption/ionization–time of flight mass spectrometry (MALDI-TOF MS) is widely used in clinical microbiology laboratories for bacterial identification but its use for detection of antimicrobial resistance (AMR) remains limited. Here, we used MALDI-TOF MS with artificial intelligence (AI) approaches to successfully predict AMR in Pseudomonas aeruginosa , a priority pathogen with complex AMR mechanisms. The highest performance was achieved for modern β-lactam/β-lactamase inhibitor drugs, namely, ceftazidime/avibactam and ceftolozane/tazobactam. For these drugs, the model demonstrated area under the receiver operating characteristic curve (AUROC) of 0.869 and 0.856, specificity of 0.925 and 0.897, and sensitivity of 0.731 and 0.714, respectively. As part of this work, we developed dynamic binning, a feature engineering technique that effectively reduces the high-dimensional feature set and has wide-ranging applicability to MALDI-TOF MS data. Compared to conventional feature engineering approaches, the dynamic binning method yielded highest performance in 7 of 10 antimicrobials. Moreover, we showcased the efficacy of transfer learning in enhancing the AUROC performance for 8 of 11 antimicrobials. By assessing the contribution of features to the model's prediction, we identified proteins that may contribute to AMR mechanisms. Our findings demonstrate the potential of combining AI with MALDI-TOF MS as a rapid AMR diagnostic tool for Pseudomonas aeruginosa . IMPORTANCE: Pseudomonas aeruginosa is a key bacterial pathogen that causes significant global morbidity and mortality. Antimicrobial resistance (AMR) emerges rapidly in P. aeruginosa and is driven by complex mechanisms. Drug-resistant P. aeruginosa is a major challenge in clinical settings due to limited treatment options. Early detection of AMR can guide antibiotic choices, improve patient outcomes, and avoid unnecessary antibiotic use. Matrix-assisted laser desorption/ionization–time of flight mass spectrometry (MALDI-TOF MS) is widely used for rapid species identification in clinical microbiology. In this study, we repurposed mass spectra generated by MALDI-TOF and used them as inputs for artificial intelligence approaches to successfully predict AMR in P. aeruginosa for multiple key antibiotic classes. This work represents an important advance toward using MALDI-TOF as a rapid AMR diagnostic for P. aeruginosa in clinical settings.

Whole-genome sequencing (WGS) data are an invaluable resource for understanding antimicrobial resistance (AMR) mechanisms. However, WGS data are high-dimensional and the lack of standardized genomic representations is a key barrier to AMR phenotype prediction. To fully explore these high-resolution data, we propose AMR-GNN, a graph deep learning-based framework that integrates multiple genomic representations with graph neural networks (GNN) to enable AMR phenotype prediction from genomic sequence data. We test AMR-GNN with Pseudomonas aeruginosa, a clinically relevant Gram-negative bacterial pathogen known for its complex AMR mechanisms. We present AMR-GNN as a proof-of-concept framework designed to address several key problems in AMR phenotype prediction with data-driven machine learning (ML) approaches, including using multiple genomic representations to enhance performance, to mitigate the influence of clonal relationships and to identify informative biomarkers to provide explainability. Follow-up validation on the largest publicly available dataset spanning both Gram-negative and Gram-positive pathogens highlights AMR-GNN’s broad applicability in detecting AMR in diverse and clinically relevant pathogen-drug combinations.

Infections caused by carbapenemase-producing organisms are a global health threat. IMP carbapenemases are one of the key drivers of these infections but little is known regarding their global epidemiology. We analyse three decades of bla _IMP gene spread using sequence data from 4556 genomes collected between 1996–2023. A total of 52 bla _IMP variants were identified across 93 bacterial species. We reconstruct the historical emergence and variant-specific epidemiologies of bla _IMP genes and showed how key variants ( bla _IMP-1 , bla _IMP-4 , bla _IMP-7 , bla _IMP-8 and bla _IMP-13 ) achieved global endemicity, while bla _IMP-26 and bla _IMP-27 became regionally endemic in Southeast Asia and North America, respectively. Dissemination was driven predominantly by horizontal gene transfer facilitated by mobile genetic elements such as class 1 integrons and insertion sequences. These elements mobilised bla _IMP genes into 52 distinct plasmid clusters (predominantly IncHI2A, IncN, IncL/M, and IncC), enabling broad inter-species transmission. Despite limited overall cross-source transmission, spillover primarily occurred between human and environmental reservoirs. Structural analysis revealed conserved IMP carbapenemase structure (mean lDDT 0.977) with convergent missense mutations at seven catalytically relevant sites. Our analysis provides a framework for understanding bla _IMP dissemination, highlighting their emergence as an important, yet under-recognised, public health threat.

Background. Heart failure with preserved ejection fraction (HFpEF) is a rising health problem with heterogeneous presentation and no evidence-based treatment. While Southeast Asia reported the highest mortality and morbidity among Asian population, little is known about the Vietnamese population, including patient characteristics, prescribing pattern and mortality rate. Methods. We conducted an observational study on 477 patients diagnosed with HFpEF from seven hospitals in Southern Vietnam from January 2019 to December 2019. Results. Mean age was 67.6 (40.9<65 years). 62.3 were female. 82.4 were diagnosed within 5 years. Dyspnea, congestion, and hypoperfusion on admission were noted in 63.9, 48.8, and 4.6 of the patients, respectively. Median ejection fraction was 63. Valvular heart disease (VHD) was the leading cause of heart failure (35.9). 78.6 had at least two comorbidities, mostly hypertension (68.6). 30.6 of the patients were hospitalized, with a median stay of 7.0 (4.010.0) days and inhospital mortality of 4.8. Older patients (65 years) were more likely to be females (OR=1.52); had multimorbid conditions (OR=3.14), including hypertension (OR=4.28), diabetes (OR=1.73), coronary artery disease (CAD) (OR=2.50), dyslipidemia (OR=1.94), and chronic kidney disease (OR=2.44); and were more frequently prescribed statin (OR=3.15). Younger individuals (<65 years) were associated with higher mineralocorticoid antagonist uptake (OR=0.52) and VHD (OR=0,40). Prescription rate for renin-angiotensin-aldosterone system inhibitor, beta blocker, mineralocorticoid antagonist, and loop diuretic was 72.5, 59.1, 43.0, and 60.6, respectively. Four phenotypes were identified, including the lean/elderly/multimorbid; congestive/metabolic; CAD-induced; and younger/atrial fibrillation (AF)/VHD. The novel phenotype younger/AF/VHD exhibited high symptom burden and poor functional capacity despite being the youngest and least multimorbid. The lean/elderly/multimorbid phenotype demonstrated the highest symptom severity and inhospital mortality. Conclusions. Our research highlights a younger, predominantly female population with high disease burden. The four novelly identified phenotypes provide contemporary and pragmatic insights into a phenotype-guided approach, exclusively targeting the Vietnamese population.

Matrix-assisted laser desorption/ionization–time of flight mass spectrometry (MALDI-TOF MS) is widely used in clinical microbiology laboratories for bacterial identification but its use for prediction of antimicrobial resistance (AMR) remains limited. Here, we used MALDI-TOF MS with artificial intelligence (AI) approaches to successfully predict AMR in Pseudomonas aeruginosa , a priority pathogen with complex AMR mechanisms. The highest performance was achieved for modern β-lactam/β-lactamase inhibitor drugs, namely ceftazidime/avibactam and ceftolozane/tazobactam, with area under the receiver operating characteristic curve (AUROC) of 0.86 and 0.87, respectively. As part of this work, we developed dynamic binning, a feature engineering technique that effectively reduces the high-dimensional feature set and has wide-ranging applicability to MALDI-TOF MS data. Compared to conventional methods, our approach yielded superior performance in 10 of 11 antimicrobials. Moreover, we showcase the efficacy of transfer learning in enhancing the performance for 7 of 11 antimicrobials. By assessing the contribution of features to the model’s prediction, we identified proteins that may contribute to AMR mechanisms. Our findings demonstrate the potential of combining AI with MALDI-TOF MS as a rapid AMR diagnostic tool for Pseudomonas aeruginosa .

Despite evidence of SGLT2 inhibitors in improving cardiovascular outcomes of heart failure with preserved ejection fraction (HFpEF), the heterogenous mechanism and characteristic multimorbidity of HFpEF require a phenotypic approach. Metabolic phenotype, one common HFpEF phenotype, has various presentations and prognoses worldwide. We aimed to identify different phenotypes of hypertensive-diabetic HFpEF, their phenotype-related outcomes, and treatment responses. The primary endpoint was time to the first event of all-cause mortality or hospitalization for heart failure (HHF). Among 233 recruited patients, 24.9% experienced primary outcomes within 12 months. A total of 3.9% was lost to follow-up. Three phenotypes were identified. Phenotype 1 (n = 126) consisted of lean, elderly females with chronic kidney disease, anemia, and concentric hypertrophy. Phenotype 2 (n = 62) included younger males with coronary artery disease. Phenotype 3 (n = 45) comprised of obese elderly with atrial fibrillation. Phenotype 1 and 2 reported higher primary outcomes than phenotype 3 (p = 0.002). Regarding treatment responses, SGLT2 inhibitor was associated with fewer primary endpoints in phenotype 1 (p = 0.003) and 2 (p = 0.001). RAAS inhibitor was associated with fewer all-cause mortality in phenotype 1 (p = 0.003). Beta blocker was associated with fewer all-cause mortality in phenotype 1 (p = 0.024) and fewer HHF in phenotype 2 (p = 0.011). Our pioneering study supports the personalized approach to optimize HFpEF management in hypertensive-diabetic patients.