Predicting mortality risk for preterm infants using deep learning models with time-series vital sign data.
Ontology highlight
ABSTRACT:
SUBMITTER: Feng J
PROVIDER: S-EPMC8280207 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Evidence that music therapy stabilises vital parameters in preterm infants is growing, but the optimal setting for therapy is still under investigation. Our study aimed to quantify the effect of physical contact during live music therapy in preterm infants born < 32 weeks' gestational age (GA) on post-therapy vital sign values. Live music therapy was delivered twice-weekly until discharge from hospital to 40 stable infants < 32 weeks' GA. Baseline and post-therapy heart rate, respiratory rate, oxygen saturation and physical contact during each session were recorded. 159 sessions were performed with, and 444 sessions without, physical contact. Descriptive and multivariable regression analyses based on directed acyclic graphs were performed. The mean GA was 28.6 ± 2.6 weeks, and 26 (65%) infants were male. Mean absolute values for heart and respiratory rates lowered during music therapy regardless of physical contact. The mean post-therapy SaO2 was higher compared to baseline values regardless of physical contact (mean differences -8.6 beats/min; -13.3 breaths/min and +2.0%). There were no clinically relevant changes in vital sign responses between therapy sessions, with or without physical contact, or adjusted post-therapy values for any of the studied vital signs. Physical contact caused better baseline and post-therapy vital sign values but did not enhance the vital sign response to music therapy. Thus, the effect of music therapy on preterm infants' vital signs is independent of physical contact and parents' presence during music therapy in the neonatal intensive care unit.
| S-EPMC9368366 | biostudies-literature
Project description:BackgroundBundling nurse caregiving interventions are promoted to minimize infant stress.PurposeTo evaluate impact of bundled nursing care and diaper change frequency on vital sign stability and skin health of preterm infants born ≤32 weeks gestation.MethodStable preterm infants on a 3-hour feeding schedule were randomly assigned to 3- vs. 6-hour diaper changes. Diapers were changed prior to 6 h if stool was present. Direct observation of bundled care events (BCE) identify caregiving activities during each BCE. Skin pH, transepidermal water loss (TEWL), and neonatal skin condition scores (NSCS) were obtained. Vital sign data (HR, RR, O2 saturation) was downloaded from bedside monitors.ResultsForty-six infants contributed to 605 BCEs. BCEs lasted on average 28 min and included nine different activities (e.g., vital signs, feeding). Significant increases in heart rate during BCEs occurred in approximately half of the observations. Among observations with a diaper change increases in heart rate during diapering occurred in over 74% of observations Infants who were awake at the beginning of BCEs had 48% lower odds of having a change in heart rate than infants who were sleeping (p = .02). There were no group differences (3- vs. 6-hour diaper change) in skin health outcomes (TEWL, pH, NSCS).ConclusionReducing diaper change frequency without stool present should be considered to minimize caregiving stress in preterm infants. Additional research should evaluate the intrusiveness and clusters of activities that significantly impact physiologic stability to better individualize the timing of routine yet intrusive activities. Clinicaltrials.gov registry # NCT03370757.
| S-EPMC8886570 | biostudies-literature
Project description:Mortality is an unfortunately common outcome of extremely and very preterm birth. Existing clinical prediction models capture mortality risk at birth but fail to account for the remainder of the hospital course. Infants born < 32 weeks gestation with complete physiologic and clinical data were included in this retrospective study. Mortality risk was quantified by conventional means (clinical factors) using the CRIB-II score and the optimal logistic regression model. A random forest (RF) model was trained using a subset of the cohort, labeling data within 6 h of death as "worry." The model was then tested using the remaining infants. A total of 275 infants were included in the study with a mean gestational age of 27 weeks, mean birth weight of 929 g, 49% female, and a mortality rate of 21%. The CRIB-II and logistic regression models had acceptable performance with sensitivities of 71% and 80% AUC scores of 0.78 and 0.84, respectively. The RF model had superior performance with a sensitivity of 88% and an AUC of 0.93. A random forest model which incorporates fixed clinical factors with the influence of aberrancies in subsequent physiology has superior performance for mortality prediction compared to conventional models.
| S-EPMC8012581 | biostudies-literature
Project description:ObjectiveTo characterize respiratory function monitor (RFM) measurements of sustained inflations and intermittent positive pressure ventilation (IPPV) delivered noninvasively to infants in the Sustained Aeration of Infant Lungs (SAIL) trial and to compare vital sign measurements between treatment arms.Study designWe analyzed RFM data from SAIL participants at 5 trial sites. We assessed tidal volumes, rates of airway obstruction, and mask leak among infants allocated to sustained inflations and IPPV, and we compared pulse rate and oxygen saturation measurements between treatment groups.ResultsAmong 70 SAIL participants (36 sustained inflations, 34 IPPV) with RFM measurements, 40 (57%) were spontaneously breathing prior to the randomized intervention. The median expiratory tidal volume of sustained inflations administered was 5.3 mL/kg (IQR 1.1-9.2). Significant mask leak occurred in 15% and airway obstruction occurred during 17% of sustained inflations. Among 34 control infants, the median expiratory tidal volume of IPPV inflations was 4.3 mL/kg (IQR 1.3-6.6). Mask leak was present in 3%, and airway obstruction was present in 17% of IPPV inflations. There were no significant differences in pulse rate or oxygen saturation measurements between groups at any point during resuscitation.ConclusionExpiratory tidal volumes of sustained inflations and IPPV inflations administered in the SAIL trial were highly variable in both treatment arms. Vital sign values were similar between groups throughout resuscitation. Sustained inflation as operationalized in the SAIL trial was not superior to IPPV to promote lung aeration after birth in this study subgroup.Trial registrationClinicaltrials.gov: NCT02139800.
| S-EPMC8604776 | biostudies-literature
Project description:In recent years, biometric radar has gained increasing attention in the field of non-touch vital sign monitoring due to its high accuracy and strong ability to detect fine-grained movements. However, most current research on biometric radar can only achieve heart rate or respiration rate monitoring in static environments, which have strict monitoring requirements and single monitoring parameters. Moreover, most studies have not applied the collected data despite their significant potential for applications. In this paper, we introduce a non-touch motion-robust vital sign monitoring system via ultra-wideband (UWB) radar based on deep learning. Nmr-VSM not only enables multi-dimensional vital sign monitoring under human motion environments but also implements cardiac anomaly detection. The design of Nmr-VSM includes three key components. Firstly, we design a UWB radar that can perform multi-dimensional vital sign monitoring, including heart rate, respiratory rate, distance, and motion status. Secondly, we collect real experimental data and analyze the impact of eight factors, such as motion status and distance, on heart rate monitoring. We then propose a deep neural network (DNN)-based heart rate data correction model that achieves high robustness in motion environments. Finally, we model the heart rate variability (HRV) of the human body and propose a convolutional neural network (CNN)-based anomaly detection model that achieves low-latency detection of heart diseases, such as ventricular tachycardia and ventricular fibrillation. Experimental results in a real environment demonstrate that Nmr-VSM can not only accurately monitor heart rate but also achieve anomaly detection with low latency.
| S-EPMC10386750 | biostudies-literature
Project description:Bronchopulmonary dysplasia (BPD) is the most prevalent and clinically significant complication of prematurity. Accurate identification of at-risk infants would enable ongoing intervention to improve outcomes. Although postnatal exposures are known to affect an infant's likelihood of developing BPD, most existing BPD prediction models do not allow risk to be evaluated at different time points, and/or are not suitable for use in ethno-diverse populations. A comprehensive approach to developing clinical prediction models avoids assumptions as to which method will yield the optimal results by testing multiple algorithms/models. We compared the performance of machine learning and logistic regression models in predicting BPD/death. Our main cohort included infants <33 weeks' gestational age (GA) admitted to a Canadian Neonatal Network site from 2016 to 2018 (n = 9,006) with all analyses repeated for the <29 weeks' GA subcohort (n = 4,246). Models were developed to predict, on days 1, 7, and 14 of admission to neonatal intensive care, the composite outcome of BPD/death prior to discharge. Ten-fold cross-validation and a 20% hold-out sample were used to measure area under the curve (AUC). Calibration intercepts and slopes were estimated by regressing the outcome on the log-odds of the predicted probabilities. The model AUCs ranged from 0.811 to 0.886. Model discrimination was lower in the <29 weeks' GA subcohort (AUCs 0.699-0.790). Several machine learning models had a suboptimal calibration intercept and/or slope (k-nearest neighbor, random forest, artificial neural network, stacking neural network ensemble). The top-performing algorithms will be used to develop multinomial models and an online risk estimator for predicting BPD severity and death that does not require information on ethnicity.
| S-EPMC8688959 | biostudies-literature
Project description:The COVID-19 pandemic has had a significant impact on public life and health worldwide, putting the world's healthcare systems at risk. The first step in stopping this outbreak is to detect the infection in its early stages, which will relieve the risk, control the outbreak's spread, and restore full functionality to the world's healthcare systems. Currently, PCR is the most prevalent diagnosis tool for COVID-19. However, chest X-ray images may play an essential role in detecting this disease, as they are successful for many other viral pneumonia diseases. Unfortunately, there are common features between COVID-19 and other viral pneumonia, and hence manual differentiation between them seems to be a critical problem and needs the aid of artificial intelligence. This research employs deep- and transfer-learning techniques to develop accurate, general, and robust models for detecting COVID-19. The developed models utilize either convolutional neural networks or transfer-learning models or hybridize them with powerful machine-learning techniques to exploit their full potential. For experimentation, we applied the proposed models to two data sets: the COVID-19 Radiography Database from Kaggle and a local data set from Asir Hospital, Abha, Saudi Arabia. The proposed models achieved promising results in detecting COVID-19 cases and discriminating them from normal and other viral pneumonia with excellent accuracy. The hybrid models extracted features from the flatten layer or the first hidden layer of the neural network and then fed these features into a classification algorithm. This approach enhanced the results further to full accuracy for binary COVID-19 classification and 97.8% for multiclass classification.
| S-EPMC8357494 | biostudies-literature
Project description:BACKGROUND:Postoperative mortality occurs in 1-2% of patients undergoing major inpatient surgery. The currently available prediction tools using summaries of intraoperative data are limited by their inability to reflect shifting risk associated with intraoperative physiological perturbations. We sought to compare similar benchmarks to a deep-learning algorithm predicting postoperative 30-day mortality. METHODS:We constructed a multipath convolutional neural network model using patient characteristics, co-morbid conditions, preoperative laboratory values, and intraoperative numerical data from patients undergoing surgery with tracheal intubation at a single medical centre. Data for 60 min prior to a randomly selected time point were utilised. Model performance was compared with a deep neural network, a random forest, a support vector machine, and a logistic regression using predetermined summary statistics of intraoperative data. RESULTS:Of 95 907 patients, 941 (1%) died within 30 days. The multipath convolutional neural network predicted postoperative 30-day mortality with an area under the receiver operating characteristic curve of 0.867 (95% confidence interval [CI]: 0.835-0.899). This was higher than that for the deep neural network (0.825; 95% CI: 0.790-0.860), random forest (0.848; 95% CI: 0.815-0.882), support vector machine (0.836; 95% CI: 0.802-870), and logistic regression (0.837; 95% CI: 0.803-0.871). CONCLUSIONS:A deep-learning time-series model improves prediction compared with models with simple summaries of intraoperative data. We have created a model that can be used in real time to detect dynamic changes in a patient's risk for postoperative mortality.
| S-EPMC6993109 | biostudies-literature
Project description:BackgroundContinuous heart rate (HR) and oxygenation (SpO2) metrics can be useful for predicting adverse events in very low birth weight (VLBW) infants. To optimize the utility of these tools, inter-site variability must be taken into account.MethodsFor VLBW infants at three neonatal intensive care units (NICUs), we analyzed the mean, standard deviation, skewness, kurtosis, and cross-correlation of electrocardiogram HR, pulse oximeter pulse rate, and SpO2. The number and durations of bradycardia and desaturation events were also measured. Twenty-two metrics were calculated hourly, and mean daily values were compared between sites.ResultsWe analyzed data from 1168 VLBW infants from birth through day 42 (35,238 infant-days). HR and SpO2 metrics were similar at the three NICUs, with mean HR rising by ~10 beats/min over the first 2 weeks and mean SpO2 remaining stable ~94% over time. The number of bradycardia events was higher at one site, and the duration of desaturations was longer at another site.ConclusionsMean HR and SpO2 were generally similar among VLBW infants at three NICUs from birth through 6 weeks of age, but bradycardia and desaturation events differed in the first 2 weeks after birth. This highlights the importance of developing predictive analytics tools at multiple sites.ImpactHR and SpO2 analytics can be useful for predicting adverse events in VLBW infants in the NICU, but inter-site differences must be taken into account in developing predictive algorithms. Although mean HR and SpO2 patterns were similar in VLBW infants at three NICUs, inter-site differences in the number of bradycardia events and duration of desaturation events were found. Inter-site differences in bradycardia and desaturation events among VLBW infants should be considered in the development of predictive algorithms.
| S-EPMC8376742 | biostudies-literature
Project description:Metal–organic frameworks (MOFs), a class of porous nanomaterials, have been widely used in gas adsorption-based applications due to their high porosities and chemical tunability. To facilitate the discovery of high-performance MOFs for different applications, a variety of machine learning models have been developed to predict the gas adsorption capacities of MOFs. Most of the predictive models are developed using traditional machine learning algorithms. However, the continuously increasing sizes of MOF datasets and the complicated relationships between MOFs and their gas adsorption capacities make deep learning a suitable candidate to handle such big data with increased computational power and accuracy. In this study, we developed models for predicting gas adsorption capacities of MOFs using two deep learning algorithms, multilayer perceptron (MLP) and long short-term memory (LSTM) networks, with a hypothetical set of about 130,000 structures of MOFs with methane and carbon dioxide adsorption data at different pressures. The models were evaluated using 10 iterations of 10-fold cross validations and 100 holdout validations. The MLP and LSTM models performed similarly with high prediction accuracy. The models for predicting gas adsorption at a higher pressure outperformed the models for predicting gas adsorption at a lower pressure. The deep learning models are more accurate than the random forest models reported in the literature, especially for predicting gas adsorption capacities at low pressures. Our results demonstrated that deep learning algorithms have a great potential to generate models that can accurately predict the gas adsorption capacities of MOFs.
| S-EPMC9565823 | biostudies-literature