The aim of this study is to develop an explainable late-onset sepsis (LOS) prediction algorithm based on continuously measured multi-channel physiological signals that can be applied to a bedside patient monitor for preterm infants in a neonatal intensive care unit (NICU). The study highlights the complementary predictive value of motion information for LOS prediction when combined with cardiorespiratory information. The algorithm uses features that contain information on heart rate variability (HRV), respiration, and motion, based on continuously measured physiological waveforms including electrocardiogram (ECG) and chest impedance (CI). In this study, 127 preterm infants were included, of whom 59 were blood-culture-proven LOS patients and 68 were control patients. Features in 24 hours before the onset of sepsis (for the LOS group), and an age-matched onset time point (for the control group) were extracted and fed into machine learning classifiers together with gestational age (GA) and birth weight. We compared the prediction performance of several well-known classifiers using features extracted from different signal channels (HRV, respiration, and motion) individually as well as their combinations. The prediction performance was evaluated using the area under the receiver-operating-characteristics curve (AUC). The best performance for LOS prediction was achieved by an XGB classifier combining features from all signal channels, with an AUC of 0.88, a positive predictive value of 0.80, and a negative predictive value of 0.83 during the 6 hours preceding LOS onset. This feasibility study demonstrates the complementary predictive value of motion information in addition to cardiorespiratory information for LOS prediction. Furthermore, visualization of how each feature measured in the individual patient impacts the algorithm decision can strengthen its interpretability. In clinical practice, it is important that clinical interventions are motivated and this visualization method can help to support the clinical decision.

Apnea of prematurity is one of the most common diagnosis in neonatal intensive care units. Apneas can be classified as central, obstructive or mixed. According to the current international standards, minimal fluctuations or absence of fluctuations in the chest impedance (CI) suggest a central apnea (CA). However, automatic detection of reduced CI fluctuations leads to a high number of central apnea-suspected events (CASEs), the majority being false alarms. We aim to improve automatic detection of CAs by using machine learning to optimize detection of CAs among CASEs. Using an optimized algorithm for automated detection, all CASEs were detected in a population of 10 premature infants developing late-onset sepsis and 10 age-matched control patients. CASEs were inspected by two clinical experts and annotated as CAs or rejections in two rounds of annotations. A total of 47 features were extracted from the ECG, CI and oxygen saturation signals considering four 30 s-long moving windows, from 30 s before to 15 s after the onset of each CASE, using a moving step size of 5 s. Consecutively, new CA detection models were developed based on logistic regression with elastic net penalty, random forest and support vector machines. Performance was evaluated using both leave-one-patient-out and 10-fold cross-validation considering the mean area under the receiver-operating-characteristic curve (AUROC). The CA detection model based on logistic regression with elastic net penalty returned the highest mean AUROC when features extracted from all four time windows were included, both using leave-one-patient-out and 10-fold cross-validation (mean AUROC of 0.88 and 0.90, respectively). Feature relevance was found to be the highest for features derived from the CI. A threshold for the false positive rate in the mean receiver-operating-characteristic curve equal to 0.3 led to a high percentage of correct detections for all CAs (78.2%) and even higher for CAs followed by a bradycardia (93.4%) and CAs followed by both a bradycardia and a desaturation (95.2%), which are more critical for the well-being of premature infants. Models based on machine learning can lead to improved CA detection with fewer false alarms.

Continuous prediction of late-onset sepsis (LOS) could be helpful for improving clinical outcomes in neonatal intensive care units (NICU). This study aimed to develop an artificial intelligence (AI) model for assisting the bedside clinicians in successfully identifying infants at risk for LOS using non-invasive vital signs monitoring. In a retrospective study from the NICU of the Máxima Medical Center in Veldhoven, the Netherlands, a total of 492 preterm infants less than 32 weeks gestation were included between July 2016 and December 2018. Data on heart rate (HR), respiratory rate (RR), and oxygen saturation (SpO₂) at 1 Hz were extracted from the patient monitor. We developed multiple AI models using 102 extracted features or raw time series to provide hourly LOS risk prediction. Shapley values were used to explain the model. For the best performing model, the effect of different vital signs and also the input type of signals on model performance was tested. To further assess the performance of applying the best performing model in a real-world clinical setting, we performed a simulation using four different alarm policies on continuous real-time predictions starting from three days after birth. A total of 51 LOS patients and 68 controls were finally included according to the patient inclusion and exclusion criteria. When tested by seven-fold cross-validations, the mean (standard deviation) area under the receiver operating characteristic curve (AUC) six hours before CRASH was 0.875 (0.072) for the best performing model, compared to the other six models with AUC ranging from 0.782 (0.089) to 0.846 (0.083). The best performing model performed only slightly worse than the model learning from raw physiological waveforms (0.886 [0.068]), successfully detecting 96.1 % of LOS patients before CRASH. When setting the expected alarm window to 24 h and using a multi-threshold alarm policy, the sensitivity metric was 71.6 %, while the positive predictive value was 9.9 %, resulting in an average of 1.15 alarms per day per patient. The proposed AI model, which learns from routinely collected vital signs, has the potential to assist clinicians in the early detection of LOS. Combined with interpretability and clinical alarm management, this model could be better translated into medical practice for future clinical implementation.

Abnormal body motion in infants may be associated with neurodevelopmental delay or critical illness. In contrast to continuous patient monitoring of the basic vitals, the body motion of infants is only determined by discrete periodic clinical observations of caregivers, leaving the infants unattended for observation for a longer time. One step to fill this gap is to introduce and compare different sensing technologies that are suitable for continuous infant body motion quantification. Therefore, we conducted this systematic review for infant body motion quantification based on the PRISMA method (Preferred Reporting Items for Systematic Reviews and Meta-Analyses). In this systematic review, we introduce and compare several sensing technologies with motion quantification in different clinical applications. We discuss the pros and cons of each sensing technology for motion quantification. Additionally, we highlight the clinical value and prospects of infant motion monitoring. Finally, we provide suggestions with specific needs in clinical practice, which can be referred by clinical users for their implementation. Our findings suggest that motion quantification can improve the performance of vital sign monitoring, and can provide clinical value to the diagnosis of complications in infants.

Dual Energy CT (DECT) has recently gained significant research interest owing to its ability to discriminate materials, and hence is widely applied in the field of nuclear safety and security inspection. With the current technological developments, DECT can be typically realized by using two sets of detectors, one for detecting lower energy X-rays and another for detecting higher energy X-rays. This makes the imaging system expensive, limiting its practical implementation. In 2009, our group performed a preliminary study on a new low-cost system design, using only a complete data set for lower energy level and a sparse data set for the higher energy level. This could significantly reduce the cost of the system, as it contained much smaller number of detector elements. Reconstruction method is the key point of this system. In the present study, we further validated this system and proposed a robust method, involving three main steps: (1) estimation of the missing data iteratively with TV constraints; (2) use the reconstruction from the complete lower energy CT data set to form an initial estimation of the projection data for higher energy level; (3) use ordered views to accelerate the computation. Numerical simulations with different number of detector elements have also been examined. The results obtained in this study demonstrate that 1 + 14% CT data is sufficient enough to provide a rather good reconstruction of both the effective atomic number and electron density distributions of the scanned object, instead of 2 sets CT data.