Background: Auscultation is a critical diagnostic feature of lung diseases, but it is subjective and challenging to measure accurately. To overcome these limitations, artificial intelligence models have been developed.Methods: In this prospective study, we aimed to compare respiratory sound feature extraction methods to develop an optimal machine learning model for detecting wheezing in children. Pediatric pulmonologists recorded and verified 103 instances of wheezing and 184 other respiratory sounds in 76 children. Various methods were used for sound feature extraction, and dimensions were reduced using t-distributed Stochastic Neighbor Embedding (t-SNE). The performance of models in wheezing detection was evaluated using a kernel support vector machine (SVM).Results: The duration of recordings in the wheezing and non-wheezing groups were 89.36 ± 39.51 ms and 63.09 ± 27.79 ms, respectively. The Mel-spectrogram, Mel-frequency Cepstral Coefficient (MFCC), and spectral contrast achieved the best expression of respiratory sounds and showed good performance in cluster classification. The SVM model using spectral contrast exhibited the best performance, with an accuracy, precision, recall, and F-1 score of 0.897, 0.800, 0.952, and 0.869, respectively.Conclusion: Mel-spectrograms, MFCC, and spectral contrast are effective for characterizing respiratory sounds in children. A machine learning model using spectral contrast demonstrated high detection performance, indicating its potential utility in ensuring accurate diagnosis of pediatric respiratory diseases.
This study investigated the clinical and genomic characteristics of Enterobacter cloacae complex (ECC) and Klebsiella aerogenes bloodstream infections (BSIs) in pediatric patients. A total of 115 BSI episodes (ECC: 86, K. aerogenes: 29) from 110 children hospitalized between 2011 and 2024 were retrospectively analyzed. Whole-genome sequencing was performed on available isolates to determine species, sequence types, and antimicrobial resistance (AMR) genes. Clinical characteristics, antibiotic usage, and outcomes were compared between groups. Patients with K. aerogenes BSI were younger and more likely to be preterm or diagnosed with urosepsis, while ECC infections were more frequently associated with hematologic malignancies. According to a multivariable analysis of the entire cohort (n = 115), K. aerogenes infection (OR [6.26], 95% CI [1.36–28.78]) and gentamicin resistance (OR [10.06], 95% CI [1.88–53.87]) were independently associated with 30-day mortality. Enterobacter hormaechei was the most common ECC species (68.4%) and exhibited the highest prevalence of AMR genes, particularly those conferring resistance to aminoglycosides, β-lactams, and trimethoprim–sulfamethoxazole. In contrast, K. aerogenes harbored few resistance genes. Multi-locus sequence typing analysis revealed high genetic diversity in both ECC and K. aerogenes, without evidence of dominant clonal expansion. Despite similarities in clinical presentation, ECC and K. aerogenes exhibit distinct age distributions, resistance profiles, and genetic diversity in pediatric BSIs. These findings underscore the importance of species-level identification and continued genomic surveillance to inform empirical antibiotic strategies and prevent the spread of resistant strains.