Early Detection Of Late-onset Sepsis Research Articles

Clinical Decision Support for Improved Neonatal Care: The Development of a Machine Learning Model for the Prediction of Late-onset Sepsis and Necrotizing Enterocolitis

To develop an artificial intelligence-based software system for predicting late-onset sepsis (LOS) and necrotizing enterocolitis (NEC) in infants admitted to the neonatal intensive care unit (NICU). Single-center, retrospective cohort study, conducted in the NICU of the Antwerp University Hospital. Continuous monitoring data of 865 preterm infants born at <32weeks gestational age, admitted to the NICU in the first week of life, were used to train an XGBoost machine learning (ML) algorithm for LOS and NEC prediction in a cross-validated setup. Afterward, the model's performance was assessed on an independent test set of 148 patients (internal validation). The ML model delivered hourly risk predictions with an overall sensitivity of 69% (142/206) for all LOS/NEC episodes and 81% (67/83) for severe LOS/NEC episodes. The model showed a median time gain of ≤10 hours (IQR, 3.1-21.0 hours), compared with historical clinical diagnosis. On the complete retrospective dataset, the ML model made 721 069 predictions, of which 9805 (1.3%) depicted a LOS/NEC probability of ≥0.15, resulting in a total alarm rate of <1 patient alarm-day per week. The model reached a similar performance on the internal validation set. Artificial intelligence technology can assist clinicians in the early detection of LOS and NEC in the NICU, which potentially can result in clinical and socioeconomic benefits. Additional studies are required to quantify further the effect of combining artificial and human intelligence on patient outcomes in the NICU.

Early Detection of Late Onset Sepsis in Extremely Preterm Infants Using Machine Learning: Towards an Early Warning System

A significant proportion of babies that are admitted to the neonatal intensive care unit (NICU) suffer from late onset sepsis (LOS). In order to prevent mortality and morbidity, the early detection of LOS is of the utmost importance. Recent works have found that the use of machine learning techniques might help detect LOS at an early stage. Some works have shown that linear methods (i.e., logistic regression) display a superior performance when predicting LOS. Nevertheless, as research on this topic is still in an early phase, it has not been ruled out that non-linear machine learning (ML) techniques can improve the predictive performance. Moreover, few studies have assessed the effect of parameters other than heart rate variability (HRV). Therefore, the current study investigates the effect of non-linear methods and assesses whether other vital parameters such as respiratory rate, perfusion index, and oxygen saturation could be of added value when predicting LOS. In contrast with the findings in the literature, it was found that non-linear methods showed a superior performance compared with linear models. In particular, it was found that random forest performed best (AUROC: 0.973), 24% better than logistic regression (AUROC: 0.782). Nevertheless, logistic regression was found to perform similarly to some non-linear models when trained with a short training window. Furthermore, when also taking training time into account, K-Nearest Neighbors was found to be the most beneficial (AUROC: 0.950). In line with the literature, we found that training the models on HRV features yielded the best results. Lastly, the results revealed that non-linear methods demonstrated a superior performance compared with linear methods when adding respiratory features to the HRV feature set, which ensured the greatest improvement in terms of AUROC score.

Development and clinical impact assessment of a machine-learning model for early prediction of late-onset sepsis

Background and aimPreterm infants are prone to neonatal infections such as late-onset sepsis (LOS). The consequences of LOS can be severe and potentially life-threatening. Unfortunately, LOS often presents with unspecific symptoms, and early screening laboratory tests have limited diagnostic value and are often late. This study aimed to build a predictive algorithm to aid doctors in the early detection of LOS in very preterm infants. MethodsIn a retrospective cohort study, all consecutively admitted preterm infants (GA ≤ 32 weeks) from 2008 until 2019 were included. They were classified as LOS or control according to blood culture results, currently the gold standard. To generate features, routine and continuously measured oxygen saturation and heart rate data with a minute-by-minute sampling rate were extracted from electronic medical records. Care was taken not to include variables indicative of existing LOS suspicion. The timing of a positive blood culture served as a proxy for LOS-onset. An equivalent timestamp was generated in gestational-age-matched control patients without a positive blood culture. Three machine learning (ML) techniques (generalized additive models, logistic regression, and XGBoost) were used to build a classification algorithm. To simulate the performance of the algorithm in clinical practice, a simulation using multiple alarm thresholds was performed on hourly predictions for the total hospitalization period. Results292 infants with LOS were matched to 1497 controls. The median gestational age before matching was 28.1 and 30.3 weeks, respectively. Evaluation of the overall discriminative power of the LR algorithm yielded an AUC of 0.73 (p < 0.05) at the moment of clinical suspicion (t = 0). In the longitudinal simulation, our algorithm detects LOS in at least 47% of the patients before clinical suspicion without exceeding the alarm fatigue threshold of 3 alarms per day. Furthermore, medical experts evaluated the algorithm as clinically relevant regarding the feature contributions in the model explanations. ConclusionsAn ML algorithm was trained for the early detection of LOS. Performance was evaluated on both prediction horizons and in a clinical impact simulation. To the best of our knowledge, our assessment of clinical impact with a retrospective simulation on longitudinal data is the most extensive in the literature on LOS prediction to date. The clinically relevant algorithm, based on routinely collected data, can potentially accelerate clinical decisions in the early detection of LOS, even with limited inputs.

Fecal Volatile Metabolomics Predict Gram-Negative Late-Onset Sepsis in Preterm Infants: A Nationwide Case-Control Study.

Early detection of late-onset sepsis (LOS) in preterm infants is crucial since timely treatment initiation is a key prognostic factor. We hypothesized that fecal volatile organic compounds (VOCs), reflecting microbiota composition and function, could serve as a non-invasive biomarker for preclinical pathogen-specific LOS detection. Fecal samples and clinical data of all preterm infants (≤30 weeks' gestation) admitted at nine neonatal intensive care units in the Netherlands and Belgium were collected daily. Samples from one to three days before LOS onset were analyzed by gas chromatography-ion mobility spectrometry (GC-IMS), a technique based on pattern recognition, and gas chromatography-time of flight-mass spectrometry (GC-TOF-MS), to identify unique metabolites. Fecal VOC profiles and metabolites from infants with LOS were compared with matched controls. Samples from 121 LOS infants and 121 matched controls were analyzed using GC-IMS, and from 34 LOS infants and 34 matched controls using GC-TOF-MS. Differences in fecal VOCs were most profound one and two days preceding Escherichia coli LOS (Area Under Curve; p-value: 0.73; p = 0.02, 0.83; p < 0.002, respectively) and two and three days before gram-negative LOS (0.81; p < 0.001, 0.85; p < 0.001, respectively). GC-TOF-MS identified pathogen-specific discriminative metabolites for LOS. This study underlines the potential for VOCs as a non-invasive preclinical diagnostic LOS biomarker.

Cardiorespiratory signature of neonatal sepsis: development and validation of prediction models in 3 NICUs.

Heart rate characteristics aid early detection of late-onset sepsis (LOS), but respiratory data contain additional signatures of illness due to infection. Predictive models using cardiorespiratory data may improve early sepsis detection. We hypothesized that heart rate (HR) and oxygenation (SpO2) data contain signatures that improve sepsis risk prediction over HR or demographics alone. We analyzed cardiorespiratory data from very low birth weight (VLBW, <1500 g) infants admitted to three NICUs. We developed and externally validated four machine learning models to predict LOS using features calculated every 10 m: mean, standard deviation, skewness, kurtosis of HR and SpO2, and cross-correlation. We compared feature importance, discrimination, calibration, and dynamic prediction across models and cohorts. We built models of demographics and HR or SpO2 features alone for comparison with HR-SpO2 models. Performance, feature importance, and calibration were similar among modeling methods. All models had favorable external validation performance. The HR-SpO2 model performed better than models using either HR or SpO2 alone. Demographics improved the discrimination of all physiologic data models but dampened dynamic performance. Cardiorespiratory signatures detect LOS in VLBW infants at 3 NICUs. Demographics risk-stratify, but predictive modeling with both HR and SpO2 features provides the best dynamic risk prediction. Heart rate characteristics aid early detection of late-onset sepsis, but respiratory data contain signatures of illness due to infection. Predictive models using both heart rate and respiratory data may improve early sepsis detection. A cardiorespiratory early warning score, analyzing heart rate from electrocardiogram or pulse oximetry with SpO2, predicts late-onset sepsis within 24 h across multiple NICUs and detects sepsis better than heart rate characteristics or demographics alone. Demographics risk-stratify, but predictive modeling with both HR and SpO2 features provides the best dynamic risk prediction. The results increase understanding of physiologic signatures of neonatal sepsis.

The role of elevated central-peripheral temperature difference in early detection of late-onset sepsis in preterm infants

AimsThe study investigated the association between clinical symptoms and late-onset sepsis (LOS) in preterm infants with the aim of identifying a non-invasive tool for the early detection of LOS. MethodsThis was a prospective study of 83 episodes of suspected LOS in 67 preterm infants. At the time LOS was suspected, we recorded a standardized set of clinical symptoms. A diagnosis of “clinical LOS” (Clin-LOS), “culture-proven LOS” (Prov-LOS) or “LOS not present” (No-LOS) was made on the basis of C-reactive protein (CrP) and blood culture results where Clin-LOS was defined as CrP>10mg/l, Prov-LOS was defined as CrP>10mg/l AND positive blood cultures, or it was established that there was no sepsis present (No-LOS). We examined univariable associations between clinical signs and LOS using odds ratio (OR) analysis and then adjusted the odds ratio (adOR) through binary regression analysis. ResultsClin-LOS was diagnosed in 20/83 episodes, 19 cases were found to have Prov-LOS. Clinical signs which had a significant association with Clin-LOS were capillary refill time >2s (OR 2.9) and decreased responsiveness (OR 5.2), whereas there was a negative association between gastric residuals and LOS (OR 0.35). However, the most marked association was found for a greater central-peripheral temperature difference (cpTD) >2°C (OR 9). In Prov-LOS an increased heart rate (OR 3.1), prolonged capillary refill time (OR 3.3) and again an increased cpTD (OR 16) had a significant association with LOS, whereas gastric residuals were negatively associated (OR 0.29). Regression analysis showed that cpTD was the most striking clinical sign associated with both Clin- (adOR 6.3) and Prov-LOS (adOR 10.5). ConclusionsProlonged capillary refill time and – more impressive – elevated cpTD were the most useful clinical symptoms for detection of LOS in preterm infants. We especially suggest using cpTD as a predictor of LOS. It is a cheap, non-invasive and readily available tool for daily routines.

Pilot Study of a New Mathematical Algorithm for Early Detection of Late-Onset Sepsis in Very Low-Birth-Weight Infants

Diagnosis of late onset sepsis (LOS) in very low birth weight (VLBW) preterm infants relies mainly on clinical suspicion, whereas prognosis depends on early initiation of antibiotic treatment. RALIS is a mathematical algorithm for early detection of LOS incorporating six vital signs measured every 2 hours. The aim of this study is to study RALIS ability to detect LOS before clinical suspicion. A total of 118 VLBW preterm infants (gestational age < 33 weeks, birth weight < 1,500 g) were enrolled in a prospective multicentered study. Vital signs were recorded prospectively up to day 21 of life in a blinded manner, with no effect on standard care. The primary end point was comparison of the rates and timing of detection of LOS between RALIS and clinical/culture evidence of LOS. Of the 2,174 monitoring days, RALIS indicated sepsis in 590 days, and LOS was positively diagnosed in 229 days. Sensitivity, specificity, positive, and negative predictive values were 74.6, 80.7, 38.8, and 95.1%, respectively. RALIS provided an indication for sepsis 3 days on the average before clinical suspicion. RALIS has a promising potential as an easy to implement noninvasive early indicator of LOS, especially for ruling out LOS in VLBW high-risk infants.

Early Detection of Late-Onset Sepsis in Very Low Birth Weight Infants

Objective: Early detection of late onset sepsis (LOS) in the NICU is hindered by the fact that suspicious symptoms are often late and nonspecific while many laboratory adjuvant tests have a poor early sensitivity. A mathematical algorithm (RALIS #1, Integralis Ltd Israel) using six routine clinical parameters measured every two hours (Body temperature, Heart Rate, Respiratory Rate, Episodes of Bradycardia and Oxygen Desaturation, Body weight) has been developed to allow for early detection of LOS. We present the results of a prospective study where this algorithm was used for early detection of LOS, in VLBW premature infants. Methods: 98 preterm infants, admitted in the neonatal intensive care unit during the period between June 2009 and January 2011, were prospectively monitored for the first 8 to 21 days of life. Gestational age was < 33 weeks and birth weight < 1700 g. 1795 days of monitoring are included. Readings from the RALIS device were compared with clinical and culture results. Results: The RALIS algorithm identified 1255 of the 1549 normal days and 238 of the 246 days with sepsis. It had a false negative rate of 3% and a false positive rate of 19%. p-value in a chi-square test was < 0.0001. As was shown in a previous retrospective study, the RALIS system detected LOS at a significant earlier age than the age at which LOS was clinically suspected. Conclusions: These results imply that RALIS may have an important role in the early detection of LOS in VLBW infants.