Other specific DSP article suggested by Editorial Board
Machine Learning-Based Prediction Model for Multidrug-Resistant Organisms Infections: Performance Evaluation and Interpretability Analysis.
DOI: 10.2147/idr.s459830
Authors: Zhao W et al
Abstract
Background: Multidrug-resistant organism (MDRO) infections pose a significant global health threat, particularly in intensive care units (ICUs), where delayed identification exacerbates clinical outcomes. Although machine learning (ML) holds promise for infection prediction, the opaque nature of complex algorithms impedes clinical adoption. This study evaluated an interpretable machine learning model incorporating SHapley Additive exPlanations (SHAP) to predict MDRO infections in ICU patients.
Methods: A retrospective cohort study was conducted on 888 ICU patients (2020-2022) from a tertiary hospital in China. Following TRIPOD guidelines, key predictors were identified using Lasso regression from a comprehensive set of clinical variables, including demographics, treatments, and laboratory data. Six machine learning algorithms-Neural Networks, Random Forests, Support Vector Machines, Logistic Regression, Decision Trees, and Gaussian Naive Bayes-were evaluated based on AUC, accuracy, and calibration curves. SHAP analysis provided both global and local interpretability.
Results: Among 825 eligible cases (375 MDRO infections), the Random Forest model exhibited the highest performance (AUC = 0.83, accuracy = 76.7%). SHAP analysis identified urinary catheterization, ventilator use, and prolonged antibiotic exposure as key modifiable risk factors. Case-level interpretation via dynamic force plots illustrated individualized risk stratification. Decision curve analysis indicated clinical utility within probability thresholds of 0.44-0.60.
Conclusion: This study establishes an interpretable prediction framework integrating RF algorithms with SHAP explainability, balancing predictive accuracy with clinical transparency. The model’s dynamic visualization capabilities support individualized risk assessment and evidence-based antimicrobial stewardship. Integration into hospital information systems with real-time dashboards could enhance early intervention strategies.
Other specific DSP article suggested by Editorial Board
Antimicrobial resistance in community-acquired enteric pathogens among children aged 10-years in low-and middle-income countries: a systematic review and meta-analysis.
Authors: Okumu N et al
Abstract
Introduction: Antimicrobial resistance (AMR) is a global health priority. This systematic review summarizes the prevalence of AMR in enteric pathogens originating from the community, specifically among ≤10-year-old children in low-and middle-income countries (LMICs). In addition, it presents the proportions of pooled resistance in Campylobacter spp., Escherichia coli, Shigella spp., and Salmonella spp. (CESS) to clinically relevant antibiotics.
Methods: Six online repositories, namely PubMed, Medline, Web of Science, Cochrane Library, CABI, and EMBASE were searched for articles published between January 2005 and September 2024. Random-effects meta-analysis models were constructed to estimate the pooled AMR proportions for CESS pathogens, and a subgroup analysis by region was also carried out.
Results: A total of 64 publications from 23 LMICs met our inclusion criteria. The pooled estimates of E. coli AMR for clinically important antibiotics were as follows: sulfamethoxazole/trimethoprim (SXT) 71% [95%CI: 57-82%]; ampicillin (AMP) 56% [95%CI: 44-67%]; ciprofloxacin (CIP) 10% [95%CI: 5-20%]; and ceftriaxone (CRO) 8% [95%CI: 2-31%]. The proportions of AMR detected in Shigella spp. were AMP 76% [95%CI: 60-87%]; nalidixic acid (NA) 9% [95%CI: 2-31%]; CIP 3% [95%CI: 0-15%]; and CRO 2% [95%CI: 0-19%]. The proportions of Salmonella spp. AMR were AMP 55% [95%CI: 35-73%] and SXT 25% [95%CI: 15-38%]. The proportions of Campylobacter spp. AMR were erythromycin (ERY) 33% [95%CI: 12-64%] and CIP 27% [95%CI: 8-61%]. There was high variability in the regional subgroup analysis, with high interstudy and regional heterogeneity I2 ≥ 75%.
Conclusion: These results shed light on drug-resistant enteric bacterial pathogens in young children, providing evidence that CESS pathogens are becoming increasingly resistant to clinically important antimicrobials. Regional differences in resistance patterns between these community isolates highlight the need for strong national and regional surveillance to detect regional variations and inform treatment and appropriate antibiotic stewardship programs. The limitations of findings include high regional variability, significant interstudy heterogeneity, and underrepresentation of certain LMICs
Other specific DSP article suggested by Editorial Board
Outcomes of Invasive Aspergillosis in Liver Transplant Recipients From an Institution Using Targeted Antifungal Prophylaxis and an Antifungal Stewardship Program.
DOI: 10.1111/tid.70046
Authors: Collis B et al
Abstract
Background: Recent evidence suggests liver transplant recipients (LiTRs) with invasive aspergillosis (IA) have lower rates of dissemination and mortality compared to historical data. However, contemporary data from other centers remain scarce. The study authors aimed to evaluate modern IA outcomes at their institution, where targeted perioperative echinocandin prophylaxis and an active antifungal stewardship program (AFSP) have been implemented.
Methods: This is a single-center retrospective analysis of patients who underwent liver transplantation between January 1, 2017 and June 30, 2022. During the study period, targeted anidulafungin perioperative prophylaxis was administered to patients considered high-risk for invasive fungal infection (IFI), and a multidisciplinary AFSP assisted with IFI diagnosis and management. Patients with proven and probable IA diagnosed post-operatively were identified using internationally accepted definitions. The primary outcomes were IA dissemination and 1-year all-cause mortality rates. Data were extracted from the electronic medical record and descriptive summary statistics were performed.
Results: Six patients (6/377, 1.6%) met the inclusion criteria. Patients with IA were significantly more likely to be colonized with multidrug-resistant Gram-negative organisms compared to those without IA (50.0% vs. 12.1%, p = 0.006). The median time to IA diagnosis was 22 days post-transplant (IQR 5-109). No cases of dissemination were observed. One-year all-cause mortality was 16.7%.
Conclusion: Consistent with contemporary data, LiTRs had lower IA dissemination and mortality rates compared to earlier studies. These improved outcomes likely reflect a combination of modern advancements in liver transplantation, and highlight two potentially modifiable interventions; targeted echinocandin prophylaxis and an AFSP. Further studies are needed to support their broader implementation.
Other specific DSP article suggested by Editorial Board
Antimicrobial Stewardship Impact on the Treatment of Intra-abdominal Infections in the Surgical Intensive Care Unit.
DOI: 10.1089/sur.2024.304
Authors: Peterson ME et al
Abstract
Background: Broad-spectrum antibiotic agents are utilized for complicated intra-abdominal infection (cIAI); however, the need for empiric methicillin-resistant Staphylococcus aureus (MRSA) coverage is not clear as the incidence of MRSA cIAI is rare.
Patients and Methods: A single-center, retrospective, pre- and post-cohort study of adults admitted to the surgical intensive care unit (SICU) with cIAI between March 1, 2021, to May 1, 2023, was conducted. Historically, the SICU utilized vancomycin for all cIAI; however, in April 2022, the preferred regimen was changed to either piperacillin-tazobactam with vancomycin added for patients with MRSA risk factors or for Enterococcus spp. coverage in cefepime- or levofloxacin-based regimens for penicillin-allergic patients. The primary outcome was number of vancomycin days of therapy (DOT) per 1,000 patient days. Categoric and continuous variables were analyzed with chi-square and Fisher exact tests.
Results: A total of 142 SICU encounters were included, 64 in the pre-cohort and 78 in the post-cohort. There was no difference in median vancomycin DOT per 1,000 patient days (14 days [interquartile range or IQR 5-21]; 16 days [IQR 8-17] p = 0.522) between the pre- and post-cohort. There was a significant reduction in the number of patients given vancomycin after the protocol change (90.6%; 76.9%, p = 0.042). A significant increase in piperacillin-tazobactam exposure was also observed (48.4%; 82.1%, p < 0.001) in the post-cohort aligning with our institutional practice change.
Conclusions: In critically ill surgical patients with cIAI, the implementation of an antimicrobial stewardship guideline did not reduce vancomycin DOT per 1,000 patient days, however, it did result in a significant reduction in vancomycin exposure.