A Machine Learning-Based Risk Score for Prediction of Infective Endocarditis Among Patients With Staphylococcus aureus Bacteremia-The SABIER Score.

BACKGROUND: Early risk assessment is needed to stratify Staphylococcus aureus infective endocarditis (SA-IE) risk among patients with S. aureus bacteremia (SAB) to guide clinical management. The objective of the current study was to develop a novel risk score that is independent of subjective clinical judgment and can be used early, at the time of blood culture positivity. METHODS: We conducted a retrospective big data analysis from territory-wide electronic data and included hospitalized patients with SAB between 2009 and 2019. We applied a random forest risk scoring model to select variables from an array of parameters, according to the statistical importance in predicting SA-IE outcome. The data were divided into derivation and validation cohorts. The areas under the curve of the receiver operating characteristic (AUCROCs) were determined. RESULTS: We identified 15 741 SAB patients, among them 658 (4.18%) had SA-IE. The AUCROC was 0.74 (95%CI 0.70-0.76), with a negative predictive value of 0.980 (95%CI 0.977-0.983). The four most discriminatory features were age, history of infective endocarditis, valvular heart disease, and community onset. CONCLUSIONS: We developed a novel risk score with performance comparable with existing scores, which can be used at the time of SAB and prior to subjective clinical judgment.

Deep learning model for prediction of extended-spectrum beta-lactamase (ESBL) production in community-onset Enterobacteriaceae bacteraemia from a high ESBL prevalence multi-centre cohort.

Adequate empirical antimicrobial coverage is instrumental in clinical management of community-onset Enterobacteriaceae bacteraemia in areas with high ESBL prevalence, while balancing the risk of carbapenem overuse and emergence of carbapenem-resistant organisms. It is unknown whether machine learning offers additional advantages to conventional statistical methods in prediction of ESBL production. To develop a validated model to predict ESBL production in Enterobacteriaceae causing community-onset bacteraemia. 5625 patients with community-onset bacteraemia caused by Escherichia coli, Klebsiella species and Proteus mirabilis during 1 January 2015-31 December 2019 from three regional hospitals in Hong Kong were included in the analysis, after exclusion of blood cultures obtained beyond 48 h of admission. The prevalence of ESBL-producing Enterobacteriaceae was 23.7% (1335/5625). Deep neural network and other machine learning algorithms were compared against conventional statistical model via multivariable logistic regression. Primary outcomes compared consisted of predictive model area under curve of receiver-operator characteristic curve (AUC), and macro-averaged F1 score. Secondary outcomes included sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Deep neural network yielded an AUC of 0.761 (95% CI 0.725-0.797) and F1 score of 0.661 (95% CI 0.633-0.689), which was superior to logistic regression (AUC 0.667 (95% CI 0.627-0.707), F1 score 0.596 (95% CI 0.567-0.625)). Deep neural network had a specificity of 91.5%, sensitivity of 37.5%, NPV of 82.5%, and PPV of 57.9%. Deep neural network is superior to logistic regression in predicting ESBL production in Enterobacteriaceae causing community-onset bacteraemia in high-ESBL prevalence area. Machine learning offers clinical utility in guiding judicious empirical antibiotics use.

Comparison of carbapenem MIC for NDM-producing Enterobacterales by different AST methods.

Introduction: This study compared the performance of MIC test strip (ETEST), automated AST card (Vitek 2) and broth microdilution (BMD) in determining carbapenem susceptibility and MIC values of NDM-producing Enterobacterales. Methods: NDM-producing Enterobacterales recovered from clinical specimens were included. The presence of blaNDM was confirmed by PCR. Identification of bacterial isolates was done by MALDI-TOF. Phenotypic susceptibility to three carbapenems (ertapenem, imipenem and meropenem) was tested by BMD, ETEST and Vitek 2. MIC values were interpreted in accordance with CLSI M100 (2022 edition). Using BMD as the reference standard, the essential agreement (EA), categorical agreement (CA), very major error (VME) and major error (ME) rates were evaluated. Results: Forty-seven NDM-producing Enterobacterales isolates were included, 44 of which were Escherichia coli. The EA of Vitek 2 was 97.9% for ertapenem, 25.5% for meropenem and 42.6% for imipenem. Using Vitek 2, there were 0% VMEs across all three carbapenems tested. The EA of ETEST was 53.2% for ertapenem, 55.3% for imipenem and 36.2% for meropenem. The rates of VMEs for ETEST were high too (ertapenem 8.5%, meropenem 36.2%, imipenem 26.1%). The MIC values obtained from Vitek 2 were consistently higher than those from BMD, while MICs from ETEST were consistently lower than those from BMD. Conclusions: The VME rate for ETEST was unacceptably high when BMD was used as the standard for comparison. Vitek 2 had acceptable EA and CA for ertapenem when BMD was used as the standard for comparison. For meropenem and imipenem, neither of the methods (ETEST, Vitek 2) showed acceptable EA and CA when compared with BMD.

Predicting antibiotic susceptibility in urinary tract infection with artificial intelligence-model performance in a multi-centre cohort.

Objective: To develop an artificial intelligence model to predict an antimicrobial susceptibility pattern in patients with urinary tract infection (UTI). Materials and methods: 26 087 adult patients with culture-proven UTI during 2015-2020 from a university teaching hospital and three community hospitals in Hong Kong were included. Cases with asymptomatic bacteriuria (absence of diagnosis code of UTI, or absence of leucocytes in urine microscopy) were excluded. Patients from 2015 to 2019 were included in the training set, while patients from the year 2020 were included as the test set.Three first-line antibiotics were chosen for prediction of susceptibility in the bacterial isolates causing UTI: namely nitrofurantoin, ciprofloxacin and amoxicillin-clavulanate. Baseline epidemiological factors, previous antimicrobial consumption, medical history and previous culture results were included as features. Logistic regression and random forest were applied to the dataset. Models were evaluated by F1-score and area under the curve-receiver operating characteristic (AUC-ROC). Results: Random forest was the best algorithm in predicting susceptibility of the three antibiotics (nitrofurantoin, amoxicillin-clavulanate and ciprofloxacin). The AUC-ROC values were 0.941, 0.939 and 0.937, respectively. The F1 scores were 0.938, 0.928 and 0.906 respectively. Conclusions: Random forest model may aid judicious empirical antibiotics use in UTI. Given the reasonable performance and accuracy, these accurate models may aid clinicians in choosing between different first-line antibiotics for UTI.

Clean clothes or dirty clothes? Outbreak investigation of carbapenem-resistant Acinetobacter baumannii related to laundry contamination through multilocus sequence typing (MLST).

OBJECTIVE: To investigate the source in an outbreak of carbapenem-resistant Acinetobacter baumannii (CRA) in a general hospital due to contamination of a laundry evaporative cooler and the laundry environment using multilocus sequence typing (MLST). METHODS: For CRA culture, clinical samples were collected from infected patients and close contacts, and environmental sampling was performed in patient surroundings and laundry facilities. MLST was used for the molecular typing of representative CRA isolates. Bacterial isolates with identical sequence types were considered epidemiologically linked and attributable to the same source. OXA genes in Acinetobacter baumannii were detected using polymerase chain reaction (PCR). RESULTS: In total, 58 patients were affected in this outbreak. The mean patient age was 75.3, and 50% were female. The most common diagnoses at admission were skin and soft-tissue infection (n = 12, 20.7%) and pneumonia (n = 12, 20.7%). OXA-23 was positive in 64.7% of isolates. A CRA isolate from the evaporative cooler in the laundry was identical to that of 11 patients across 3 wards, belonging to ST345. Isolates from 3 laundry linen racks were identical to those of 7 patients from 3 wards, classified as ST1145. Isolates found on another linen rack and a pajama shelf were identical to isolates from 3 other patients from 2 wards, belonging to ST2207. There was no significant difference between sequence type distributions of clinical and environmental isolates (P = .12), indicating high likelihood of CRA originating from the same source. CONCLUSIONS: MLST confirmed that contamination of the laundry evaporative cooler and surrounding environment caused a polyclonal CRA hospital outbreak. Hospital laundry is an important area for infection control and outbreak investigations of CRA.

Improving mobility in the intensive care unit with a protocolized, early mobilization program: observations of a single center before-and-after the implementation of a multidisciplinary program.

BACKGROUND: Early intensive care unit (ICU) protocolized rehabilitative programs have been described previously, yet with differing starting time points and mostly on mechanically ventilated patients. We extended the concept to all admitted ICU patients and investigate the efficacy of early mobilization in improving mobility of the critically ill, address issues surrounding the timing and intensity of an early rehabilitative program. METHODS: Prospective cohorts of patients admitted consecutively before-and-after (control, n=92; intervention, n=90) the introduction of an early mobilization program in a single center, general hospital ICU. Improvement in mobility as assessed by ICU mobility score, on ICU admission and upon ICU discharge, was measured as a primary outcome. RESULTS: Those receiving early mobilization in the intensive care unit had higher ICU mobility score (2.63; 95% confidence interval, 0.65-4.61; P<0.001) upon discharge from the intensive care, with earlier out of bed mobilization on day 5 compared to the control group of day 21 (P<0.001). No differences were found in terms of mortality, intensive care hospitalization and subsequent hospitalization duration after discharge from ICU. CONCLUSIONS: Here, we report that improvement in mobility score earlier in the course of intensive care hospitalization with the introduction of a protocolized early rehabilitative program.

Does expert opinion match the operational definition of the Lupus Low Disease Activity State (LLDAS)? A case-based construct validity study.

OBJECTIVE: To evaluate the construct validity of the Lupus Low Disease Activity State (LLDAS), a treatment target in systemic lupus erythematosus (SLE). METHODS: Fifty SLE case summaries based on real patients were prepared and assessed independently for meeting the operational definition of LLDAS. Fifty international rheumatologists with expertise in SLE, but with no prior involvement in the LLDAS project, responded to a survey in which they were asked to categorize the disease activity state of each case as remission, low, moderate, or high. Agreement between expert opinion and LLDAS was assessed using Cohen's kappa. RESULTS: Overall agreement between expert opinion and the operational definition of LLDAS was 77.96% (95% CI: 76.34-79.58%), with a Cohen's kappa of 0.57 (95% CI: 0.55-0.61). Of the cases (22 of 50) that fulfilled the operational definition of LLDAS, only 5.34% (59 of 22 × 50) of responses classified the cases as moderate/high activity. Of the cases that did not fulfill the operational definition of LLDAS (28 of 50), 35.14% (492 of 28 × 50) of responses classified the cases as remission/low activity. Common reasons for discordance were assignment to remission/low activity of cases with higher corticosteroid doses than defined in LLDAS (prednisolone ≤ 7.5mg) or with SLEDAI-2K >4 due to serological activity (high anti-dsDNA antibody and/or low complement). CONCLUSIONS: LLDAS has good construct validity with high overall agreement between the operational definition of LLDAS and expert opinion. Discordance of results suggests that the operational definition of LLDAS is more stringent than expert opinion at defining a low disease activity state.