RESUMO
BACKGROUND AND OBJECTIVE: Critically ill children may suffer from impaired neurocognitive functions years after ICU (intensive care unit) discharge. To assess neurocognitive functions, these children are subjected to a fixed sequence of tests. Undergoing all tests is, however, arduous for former pediatric ICU patients, resulting in interrupted evaluations where several neurocognitive deficiencies remain undetected. As a solution, we propose using machine learning to predict the optimal order of tests for each child, reducing the number of tests required to identify the most severe neurocognitive deficiencies. METHODS: We have compared the current clinical approach against several machine learning methods, mainly multi-target regression and label ranking methods. We have also proposed a new method that builds several multi-target predictive models and combines the outputs into a ranking that prioritizes the worse neurocognitive outcomes. We used data available at discharge, from children who participated in the PEPaNIC-RCT trial (ClinicalTrials.gov-NCT01536275), as well as data from a 2-year follow-up study. The institutional review boards at each participating site have also approved this follow-up study (ML8052; NL49708.078; Pro00038098). RESULTS: Our proposed method managed to outperform other machine learning methods and also the current clinical practice. Precisely, our method reaches approximately 80% precision when considering top-4 outcomes, in comparison to 65% and 78% obtained by the current clinical practice and the state-of-the-art method in label ranking, respectively. CONCLUSIONS: Our experiments demonstrated that machine learning can be competitive or even superior to the current testing order employed in clinical practice, suggesting that our model can be used to severely reduce the number of tests necessary for each child. Moreover, the results indicate that possible long-term adverse outcomes are already predictable as early as at ICU discharge. Thus, our work can be seen as the first step to allow more personalized follow-up after ICU discharge leading to preventive care rather than curative.
Assuntos
Unidades de Terapia Intensiva Pediátrica , Aprendizado de Máquina , Humanos , Criança , Masculino , Feminino , Pré-Escolar , Estado Terminal , Seguimentos , Alta do PacienteRESUMO
In critically ill adults, high plasma cortisol in face of low ACTH coincides with high pro-opiomelanocortin (POMC) levels. Glucocorticoids further lower ACTH without affecting POMC. We hypothesized that in pediatric cardiac surgery-induced critical illness, plasma POMC is elevated, plasma ACTH transiently rises intraoperatively but becomes suppressed post-operatively, and glucocorticoid administration amplifies this phenotype. From 53 patients (0-36 months), plasma was obtained pre-operatively, intraoperatively and on post-operative day 1 and 2. Plasma was also collected from 24 healthy children. In patients, POMC was supra-normal pre-operatively (p<0.0001) but no longer thereafter (p<0.05). ACTH was never high in patients. While in glucocorticoid-naive patients ACTH became suppressed by post-operative day 1 (p<0.0001), glucocorticoid-treated patients had suppressed ACTH already intraoperatively (p≤0.0001). Pre-operatively high POMC, not accompanied by increased plasma ACTH, suggests a centrally-activated HPA-axis with reduced pituitary processing of POMC into ACTH. Increasing systemic glucocorticoid availability with glucocorticoid treatment accelerated the suppression of plasma ACTH.
RESUMO
BACKGROUND: Critically ill children suffer from impaired physical/neurocognitive development 2 years later. Glucocorticoid treatment alters DNA methylation within the hypothalamus-pituitary-adrenal (HPA) axis which may impair normal brain development, cognition and behaviour. We tested the hypothesis that paediatric-intensive-care-unit (PICU) patients, sex- and age-dependently, show long-term abnormal DNA methylation within the HPA-axis layers, possibly aggravated by glucocorticoid treatment in the PICU, which may contribute to the long-term developmental impairments. RESULTS: In a pre-planned secondary analysis of the multicentre PEPaNIC-RCT and its 2-year follow-up, we identified differentially methylated positions and differentially methylated regions within HPA-axis genes in buccal mucosa DNA from 818 former PICU patients 2 years after PICU admission (n = 608 no glucocorticoid treatment; n = 210 glucocorticoid treatment) versus 392 healthy children and assessed interaction with sex and age, role of glucocorticoid treatment in the PICU and associations with long-term developmental impairments. Adjusting for technical variation and baseline risk factors and correcting for multiple testing (false discovery rate < 0.05), former PICU patients showed abnormal DNA methylation of 26 CpG sites (within CRHR1, POMC, MC2R, NR3C1, FKBP5, HSD11B1, SRD5A1, AKR1D1, DUSP1, TSC22D3 and TNF) and three DNA regions (within AVP, TSC22D3 and TNF) that were mostly hypomethylated. These abnormalities were sex-independent and only partially age-dependent. Abnormal methylation of three CpG sites within FKBP5 and one CpG site within SRD5A1 and AKR1D1 was partly attributable to glucocorticoid treatment during PICU stay. Finally, abnormal methylation within FKBP5 and AKR1D1 was most robustly associated with long-term impaired development. CONCLUSIONS: Two years after critical illness in children, abnormal methylation within HPA-axis genes was present, predominantly within FKBP5 and AKR1D1, partly attributable to glucocorticoid treatment in the PICU, and explaining part of the long-term developmental impairments. These data call for caution regarding liberal glucocorticoid use in the PICU.