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.

Neonatal Intensive-Care Unit Graduates Show Persistent Difficulties in an Intra-Dimensional Shift Card Sort.

Neonatal intensive-care unit (NICU) graduates, a group at risk for attention problems and ADHD, performed an intra-dimensional shift card sort at 34, 42, 51, and 60 months to assess executive function and to examine effects of individual risk factors. In the 'silly' game, children sorted cards (airplanes and dogs) so they were not the same as targets. In the 'same' game they did the opposite. Performance on the 'silly' game was poor, especially when it was presented first. Success in following 'silly' game rules improved with age, and was significantly linked to maternal education and birth weight for gestational age, a measure of intrauterine stress. Degree of CNS injury differentiated children who completed the task from children who did not, and also affected the need to repeat instructions in the 'same' game. These results confirm an increased likelihood of impairments in executive function during preschool years in NICU graduates.

The development of selective attention and inhibition in NICU graduates during the preschool years.

Neonatal intensive care unit (NICU) graduates have a higher incidence of attention problems including attention deficit hyperactivity disorder (ADHD). Thus, we examined the effect of risk factors (birth weight (BW), central nervous system (CNS) injury, gender, maternal education) on attention/inhibition during reaction time, continuous performance and Go/No-Go tasks at 42, 51, and 60 months (n = 271). Very low BW NICU graduates (<1,500 g) performed worse than typical BW ones (>2,500 g), displaying poorer target/non-target discrimination. Males responded faster than females, but made more false alarms and random responses. Despite short duration tasks, attention waned. Performance improved with age, but even at 60 months children had difficulty inhibiting random responding.

Auditory brainstem evoked responses in newborns with Down syndrome.

Auditory brainstem evoked responses (ABRs) were compared in 15 newborns with Down syndrome and 15 sex-, age-, and weight-matched control newborns. Participants had normal ABRs based upon values specific to 32- to 42-weeks postconceptional age. Although Wave III and Wave V component latencies and the Wave I-III interpeak latency (IPL) were shorter in ABRs of infants with Down syndrome, the Wave III-V IPL was not, pointing to anomalies in the lower rather than upper brainstem auditory pathways. Shorter Down syndrome ABR latencies have been reported at many ages. Extending these findings to newborns suggests that the underlying basis for this develops prenatally. ABR patterns in infants with Down syndrome were similar to reports for intrauterine growth restricted newborns.

Dual-task processing as a measure of executive function: a comparison between adults with Williams and Down syndromes.

Behavioral phenotypes of individuals with Williams syndrome and individuals with Down syndrome have been contrasted in relation to short-term memory. People with Down syndrome are stronger visuospatially and those with Williams syndrome are stronger verbally. We examined short-term memory, then explored whether dual-task processing further characterized behavioral phenotypes in 53 older adults with Down syndrome, 10 with Williams syndrome, and 39 controls. Short-term memory profiles generally conformed to those of younger individuals. Pegs placement and number repetition were performed singly and simultaneously. There were no etiology group performance differences on single tasks. During concurrent processing, all groups maintained single-task performance on pegs, but declined on number repetition. However, participants with Down syndrome declined more, suggesting relatively greater weakness in the dual-task processing component of executive function for this group.

Age-associated memory changes in adults with williams syndrome.

Age-associated changes on measures of episodic and working memory were examined in 15 adults with Williams Syndrome (WS; M age = 48.3 years, SD = 14.7; M IQ = 62.9, SD = 8.5) and their performance was compared to that of 33 adults with mental retardation (MR) with unspecified etiologies (M age = 54.2 years, SD = 8.9; M IQ = 61.7, SD = 6.5). Among the group with WS, older adults were significantly poorer than younger adults on the free recall task, a measure of episodic memory. Although this finding is consistent with normal aging, it occurred at a chronologically early age in adults with WS and was not found in their peers with unspecified MR. Although both groups showed small declines with age on a backward digit span task, a measure of working memory, for the group with WS the rate of decline on backward digit span was slower as compared to their performance on the free recall task. The findings from this study indicate a chronologically early and precipitous age-associated decrease in long-term, episodic memory in adults with WS.