Predicting in-hospital length of stay for very-low-birth-weight preterm infants using machine learning techniques

Benefits and Risks of Preprepared Parenteral Nutrition for Early Amino Acid Administration in Premature Infants with Very Low Birth Weight

Purpose

Administering early parenteral amino acids to very low birth weight (VLBW) premature infants (birth body weight [BBW]<1,500 g) is challenging due to factors such as holidays, cost, and access to sterile compounding facilities. Using advance-prepared parenteral nutrition (PN) may address this issue and should be evaluated for its safety and potential benefits.

Methods

We extracted data from medical records collected between July 2015 and August 2019. VLBW infants received PN for at least seven days and were split into two groups: the traditional group (n=30), which initially received a glucose solution and then PN on workdays, and the pre-preparation group (n=16), which received advance-prepared PN immediately upon admission to the neonatal intensive care unit.

Results

The median BBWs of the traditional and pre-preparation groups were 1,180.0 vs. 1,210.0 g. In the initial two days, the pre-preparation group had a significantly higher amino acid intake (2.23 and 2.24 g/kg/d) than the traditional group (0 and 1.78 g/kg/d). The pre-preparation group exhibited greater head circumference growth ratio relative to birth (7th day: 1.21% vs. -3.57%, p=0.014; 21st day: 7.71% vs. 3.31%, p=0.017). No significant differences in metabolic tolerance were observed.

Conclusion

Advanced preparation of PN can be safely implemented in VLBW preterm infants, offering advantages such as early, higher amino acid intake and improved head circumference growth within the first 21 days post-birth. This strategy may serve as a viable alternative in settings where immediate provision of sterile compounding facilities is challenging.

Postpartum length of hospital stay among obstetric patients in Ibadan, Nigeria

BACKGROUND

Postpartum Length of hospital stay (PLOHS) is an essential indicator of the quality of maternal and perinatal healthcare services. Identifying the factors associated with PLOHS will inform targeted interventions to reduce unnecessary hospitalisations and improve patient outcomes after childbirth. Therefore, we assessed the length of hospital stay after birth and the associated factors in Ibadan, Nigeria.

METHODS

We used the Ibadan Pregnancy Cohort Study (IbPCS) data, and examined the 1057 women who had information on PLOHS the mode of delivery [spontaneous vagina delivery (SVD) or caesarean section (C/S)]. The outcome variable was PLOHS, which was described as the time interval between the delivery of the infant and discharge from the health facility. PLOHS was prolonged if > 24 h for SVD and > 96 h for C/S, but normal if otherwise. Data were analysed using descriptive statistics, a chi-square test, and modified Poisson regression. The prevalence-risk ratio (PR) and 95% confidence interval (CI) are presented at the 5% significance level.

RESULTS

The mean maternal age was (30.0 ± 5.2) years. Overall, the mean PLOHS for the study population was 2.6 (95% CI: 2.4-2.7) days. The average PLOHS for women who had vaginal deliveries was 1.7 (95%CI: 1.5-1.9) days, whereas those who had caesarean deliveries had an average LOHS of 4.4 (95%CI: 4.1-4.6) days. About a third had prolonged PLOHS: SVD 229 (32.1%) and C/S 108 (31.5%). Factors associated with prolonged PLOHS with SVD, were high income (aPR = 1.77; CI: 1.13, 2.79), frequent ANC visits (> 4) (aPR = 2.26; CI: 1.32, 3.87), and antenatal admission: (aPR = 1.88; CI: 1.15, 3.07). For C/S: maternal age > 35 years (aPR = 1.59; CI: 1.02, 2.47) and hypertensive disease in pregnancy (aPR = 0.61 ; CI: 0.38, 0.99) were associated with prolonged PLOHS.

CONCLUSION

The prolonged postpartum length of hospital stay was common among our study participants occurring in about a third of the women irrespective of the mode of delivery. Maternal income, advanced maternal age, ANC related issues were predisposing factors for prolonged LOHS. Further research is required to examine providers' perspectives on PLOHS among obstetric patients in our setting.

Association between Intermittent Hypoxemia and NICU Length of Stay in Preterm Infants

INTRODUCTION

Length of hospitalization varies widely in preterm infants and can be affected by multiple maternal and neonatal factors including respiratory instability. Therefore, we aimed to determine the association between postnatal intermittent hypoxemia (IH) and prolonged hospitalization.

METHODS

This prospective single-center cohort study followed infants born at <31 weeks of gestational age through 2 years corrected age with detailed oxygen saturation data captured from days 7 to 30 of age.

RESULTS

51/164 (31%) of infants were discharged after 400/7 weeks of corrected gestational age (CGA). A greater average daily number of IH events (OR per 10 events/day 1.33 [95% CI 1.03-1.72]), duration of events (OR per minute 1.14 [1.07-1.21]), and percent time with oxygen saturation <80% (OR per percent 1.88 [1.25-2.85]) on days 7-30 of age were all significantly associated with prolonged hospitalization past 400/7 weeks CGA. In survival analyses, infants with a greater average daily number of IH events (HR per 10 events/day 0.89 [0.81-0.98]), percent time with oxygen saturation <80% (HR per percent 0.79 [0.67-0.94]), and duration of events (HR per minute 0.93 [0.91-0.95]) on days 7-30 of age all had significantly lower probability of earlier discharge. In addition, there was a significant interaction with gestational age; the association between IH and prolonged hospitalization was stronger in more mature infants (p = 0.024).

CONCLUSIONS

Physiological instability on days 7-30 of age, as manifested by IH, is significantly associated with prolonged hospitalization. IH likely represents both a marker of initial severity of illness and the beginning of biological cascades, leading to prematurity-associated morbidities.

Predicting mental and psychomotor delay in very pre-term infants using machine learning

BACKGROUND

Very preterm infants are at elevated risk for neurodevelopmental delays. Earlier prediction of delays allows timelier intervention and improved outcomes. Machine learning (ML) was used to predict mental and psychomotor delay at 25 months.

METHODS

We applied RandomForest classifier to data from 1109 very preterm infants recruited over 20 years. ML selected key predictors from 52 perinatal and 16 longitudinal variables (1-22 mo assessments). SHapley Additive exPlanations provided model interpretability.

RESULTS

Balanced accuracy with perinatal variables was 62%/61% (mental/psychomotor). Top predictors of mental and psychomotor delay overlapped and included: birth year, days in hospital, antenatal MgSO4, days intubated, birth weight, abnormal cranial ultrasound, gestational age, mom's age and education, and intrauterine growth restriction. Highest balanced accuracy was achieved with 19-month follow-up scores and perinatal variables (72%/73%).

CONCLUSIONS

Combining perinatal and longitudinal data, ML modeling predicted 24 month mental/psychomotor delay in very preterm infants ½ year early, allowing intervention to start that much sooner. Modeling using only perinatal features fell short of clinical application. Birth year's importance reflected a linear decline in predicting delay as birth year became more recent.

IMPACT

Combining perinatal and longitudinal data, ML modeling was able to predict 24 month mental/psychomotor delay in very preterm infants ½ year early (25% of their lives) potentially advancing implementation of intervention services. Although cognitive/verbal and fine/gross motor delays require separate interventions, in very preterm infants there is substantial overlap in the risk factors that can be used to predict these delays. Birth year has an important effect on ML prediction of delay in very preterm infants, with those born more recently (1989-2009) being increasing less likely to be delayed, perhaps reflecting advances in medical practice.

Opportunities and Challenges of Prognostic Models for Extremely Preterm Infants

Predicting the short- and long-term outcomes of extremely preterm infants remains a challenge. Multivariable prognostic models might be valuable tools for clinicians, parents, and policymakers for providing accurate outcome estimates. In this perspective, we discuss the opportunities and challenges of using prognostic models in extremely preterm infants at population and individual levels. At a population level, these models could support the development of guidelines for decisions about treatment limits and may support policy processes such as benchmarking and resource allocation. At an individual level, these models may enhance prenatal counselling conversations by considering multiple variables and improving transparency about expected outcomes. Furthermore, they may improve consistency in projections shared with parents. For the development of prognostic models, we discuss important considerations such as predictor and outcome measure selection, clinical impact assessment, and generalizability. Lastly, future recommendations for developing and using prognostic models are suggested. Importantly, the purpose of a prognostic model should be clearly defined, and integrating these models into prenatal counselling requires thoughtful consideration.

Infant Low Birth Weight Prediction Using Graph Embedding Features

Low Birth weight (LBW) infants pose a serious public health concern worldwide in both the short and long term for infants and their mothers. Infant weight prediction prior to birth can help to identify risk factors and reduce the risk of infant morbidity and mortality. Although many Machine Learning (ML) algorithms have been proposed for LBW prediction using maternal features and produced considerable model performance, their performance needs to be improved so that they can be adapted in real-world clinical settings. Existing algorithms used for LBW classification often fail to capture structural information from the tabular dataset of patients with different complications. Therefore, to improve the LBW classification performance, we propose a solution by transforming the tabular data into a knowledge graph with the aim that patients from the same class (normal or LBW) exhibit similar patterns in the graphs. To achieve this, several features related to each node are extracted such as node embedding using node2vec algorithm, node degree, node similarity, nearest neighbors, etc. Our method is evaluated on a real-life dataset obtained from a large cohort study in the United Arab Emirates which contains data from 3453 patients. Multiple experiments were performed using the seven most commonly used ML models on the original dataset, graph features, and a combination of features, respectively. Experimental results show that our proposed method achieved the best performance with an area under the curve of 0.834 which is over 6% improvement compared to using the original risk factors without transforming them into knowledge graphs. Furthermore, we provide the clinical relevance of the proposed model that are important for the model to be adapted in clinical settings.