Risk Factors for Perinatal Arterial Ischemic Stroke: A Machine Learning Approach.

BACKGROUND AND OBJECTIVES: Perinatal arterial ischemic stroke (PAIS) is a focal vascular brain injury presumed to occur between the fetal period and the first 28 days of life. It is the leading cause of hemiparetic cerebral palsy. Multiple maternal, intrapartum, delivery, and fetal factors have been associated with PAIS, but studies are limited by modest sample sizes and complex interactions between factors. Machine learning approaches use large and complex data sets to enable unbiased identification of clinical predictors but have not yet been applied to PAIS. We combined large PAIS data sets and used machine learning methods to identify clinical PAIS factors and compare this data-driven approach with previously described literature-driven clinical prediction models. METHODS: Common data elements from 3 registries with patients with PAIS, the Alberta Perinatal Stroke Project, Canadian Cerebral Palsy Registry, International Pediatric Stroke Study, and a longitudinal cohort of healthy controls (Alberta Pregnancy Outcomes and Nutrition Study), were used to identify potential predictors of PAIS. Inclusion criteria were term birth and idiopathic PAIS (absence of primary causative medical condition). Data including maternal/pregnancy, intrapartum, and neonatal factors were collected between January 2003 and March 2020. Common data elements were entered into a validated random forest machine learning pipeline to identify the highest predictive features and develop a predictive model. Univariable analyses were completed post hoc to assess the relationship between each predictor and outcome. RESULTS: A machine learning model was developed using data from 2,571 neonates, including 527 cases (20%) and 2,044 controls (80%). With a mean of 21 features selected, the random forest machine learning approach predicted the outcome with approximately 86.5% balanced accuracy. Factors that were selected a priori through literature-driven variable selection that were also identified as most important by the machine learning model were maternal age, recreational substance exposure, tobacco exposure, intrapartum maternal fever, and low Apgar score at 5 minutes. Additional variables identified through machine learning included in utero alcohol exposure, infertility, miscarriage, primigravida, meconium, spontaneous vaginal delivery, neonatal head circumference, and 1-minute Apgar score. Overall, the machine learning model performed better (area under the curve [AUC] 0.93) than the literature-driven model (AUC 0.73). DISCUSSION: Machine learning may be an alternative, unbiased method to identify clinical predictors associated with PAIS. Identification of previously suggested and novel clinical factors requires cautious interpretation but supports the multifactorial nature of PAIS pathophysiology. Our results suggest that identification of neonates at risk of PAIS is possible.

Relationship Between Neonatal Brain Injury and Objective Measures of Head Trauma: A Case-Control Study.

BACKGROUND AND OBJECTIVES: Neonatal brain injury is a common and devastating diagnosis conferring lifelong challenges for children and families. The role of mechanical forces applied to the head, often referred to as "birth trauma," are often considered although evidence for this association is lacking. The objective of this study was to investigate the association between common types of neonatal brain injury and scalp swelling using a novel method to quantify scalp swelling as an unbiased proxy for mechanical forces applied to the head. METHODS: Case-control study using population-based, prospectively collected tertiary care center databases and healthy controls from the Human Connectome Development Project. Included were infants born 32-42 weeks gestational age and MRI in the first 9 days. Outcomes categories included healthy neonates, hypoxic ischemic encephalopathy (HIE) with or without brain injury, or stroke (ischemic or hemorrhagic). Volume of scalp swelling was objectively quantified by a novel imaging method blinded to brain injury. Variables included mode of delivery and use of instrumentation. Statistical tests included Kruskal-Wallis test, chi square, and multivariable and multinomial logistic regression. RESULTS: There were 309 infants included (55% male): 72 healthy controls, 77 HIE without brain injury on MRI, 78 HIE with brain injury, and 82 with stroke (60 ischemic, 22 hemorrhagic). Scalp swelling was present in 126 (40.8%, 95% confidence interval [CI] 35.2%-46.5%) with no difference in proportions between outcome groups. Compared to healthy controls, median volume was higher in those with HIE without brain injury (17.5 mL, 95% CI 6.8-28.2), HIE with brain injury (12.1 mL, 95% CI 5.5-18.6), but not ischemic stroke (4.7 mL, 95% CI -1.2-10.6) nor hemorrhagic stroke (8.3 mL, 95% CI -2.2-18.8). Scalp swelling was associated with instrumented delivery (OR 2.1, 95% CI 1.0-4.1), but not associated with increased odds of brain injury in those with HIE (OR 1.5, 95% CI 0.76-3.30). Scalp swelling measures were highly reliable (ICC = 0.97). DISCUSSION: "Birth trauma" quantified by scalp swelling volume was more common in infants with difficult deliveries, but not associated with greater odds of brain injury due to hypoxia or stroke. These results may help parents and practitioners to dissociate the appearance of trauma with the risk of brain injury.

Nonlinear age effects in tactile processing from early childhood to adulthood.

BACKGROUND: Tactile processing plays a pivotal role in the early stages of human development; however, little is known about tactile function in young children. An understanding of how tactile processing changes with age from early childhood to adulthood is fundamental in understanding altered tactile experiences in neurodevelopmental disorders, such as autism spectrum disorder. METHODS: In this cross-sectional study, 142 children and adults aged 3-23 years completed a vibrotactile testing battery consisting of 5 tasks, which rely on different cortical and cognitive mechanisms. The battery was designed to be suitable for testing in young children to investigate how tactile processing changes from early childhood to adulthood. RESULTS: Our results suggest a pattern of rapid, age-related changes in tactile processing toward lower discrimination thresholds (lower discrimination thresholds = greater sensitivity) across early childhood, though we acknowledge limitations with cross-sectional data. Differences in the rate of change across tasks were observed, with tactile performance reaching adult-like levels at a younger age on some tasks compared to others. CONCLUSIONS: While it is known that early childhood is a period of profound development including tactile processing, our data provides evidence for subtle differences in the developmental rate of the various underlying cortical, physical, and cognitive processes. Further, we are the first to show the feasibility of vibrotactile testing in early childhood (<6 years). The results of this work provide estimates of age-related differences in performance, which could have important implications as a reference for investigating altered tactile processing in developmental disorders.

Bilateral actigraphic quantification of upper extremity movement in hemiparetic children with perinatal stroke: a case control study.

BACKGROUND: Hemiparetic cerebral palsy impacts millions of people worldwide. Assessment of bilateral motor function in real life remains a major challenge. We evaluated quantification of upper extremity movement in hemiparetic children using bilateral actigraphy. We hypothesized that movement asymmetry correlates with standard motor outcome measures. METHODS: Hemiparetic and control participants wore bilateral wrist Actiwatch2 (Philips) for 48 h with movement counts recorded in 15-s intervals. The primary outcome was a novel statistic of movement asymmetry, the Actigraphic Movement Asymmetry Index (AMAI). Relationships between AMAI and standard motor outcomes (Assisting Hand Assessment, Melbourne Assessment, and Box and Block Test [BB]) were explored with Pearson or Spearman correlation. RESULTS: 30 stroke (mean 11 years 2 months (3 years 10 months); 13 female, 17 male) and 23 control (mean 11 years 1 month (4 years 5 months); 8 female, 15 male) were enrolled. Stroke participants demonstrated higher asymmetry. Correlations between AMAI and standard tests were moderate and strongest during sleep (BB: r = 0.68, p < 0.01). CONCLUSIONS: Standard tests may not reflect the extent of movement asymmetry during daily life in hemiparetic children. Bilateral actigraphy may be a valuable complementary tool for measuring arm movement, potentially enabling improved evaluation of therapies with a focus on child participation.

Effects of Transcranial Direct Current Stimulation and High-Definition Transcranial Direct Current Stimulation Enhanced Motor Learning on Robotic Transcranial Magnetic Stimulation Motor Maps in Children.

Introduction: Conventional transcranial direct current stimulation (tDCS) and high-definition tDCS (HD-tDCS) may improve motor learning in children. Mechanisms are not understood. Neuronavigated robotic transcranial magnetic stimulation (TMS) can produce individualised maps of primary motor cortex (M1) topography. We aimed to determine the effects of tDCS- and HD-tDCS-enhanced motor learning on motor maps. Methods: Typically developing children aged 12-18 years were randomised to right M1 anodal tDCS, HD-tDCS, or Sham during training of their left-hand on the Purdue Pegboard Task (PPT) over 5 days. Bilateral motor mapping was performed at baseline (pre), day 5 (post), and 6-weeks retention time (RT). Primary muscle was the first dorsal interosseous (FDI) with secondary muscles of abductor pollicis brevis (APB) and adductor digiti minimi (ADM). Primary mapping outcomes were volume (mm2/mV) and area (mm2). Secondary outcomes were centre of gravity (COG, mm) and hotspot magnitude (mV). Linear mixed-effects modelling was employed to investigate effects of time and stimulation type (tDCS, HD-tDCS, Sham) on motor map characteristics. Results: Twenty-four right-handed participants (median age 15.5 years, 52% female) completed the study with no serious adverse events or dropouts. Quality maps could not be obtained in two participants. No effect of time or group were observed on map area or volume. LFDI COG (mm) differed in the medial-lateral plane (x-axis) between tDCS and Sham (p = 0.038) from pre-to-post mapping sessions. Shifts in map COG were also observed for secondary left-hand muscles. Map metrics did not correlate with behavioural changes. Conclusion: Robotic TMS mapping can safely assess motor cortex neurophysiology in children undergoing motor learning and neuromodulation interventions. Large effects on map area and volume were not observed while changes in COG may occur. Larger controlled studies are required to understand the role of motor maps in interventional neuroplasticity in children.

Robotic transcranial magnetic stimulation motor maps and hand function in adolescents.

INTRODUCTION: Transcranial magnetic stimulation (TMS) motor mapping can characterize the neurophysiology of the motor system. Limitations including human error and the challenges of pediatric populations may be overcome by emerging robotic systems. We aimed to show that neuronavigated robotic motor mapping in adolescents could efficiently produce discrete maps of individual upper extremity muscles, the characteristics of which would correlate with motor behavior. METHODS: Typically developing adolescents (TDA) underwent neuronavigated robotic TMS mapping of bilateral motor cortex. Representative maps of first dorsal interosseous (FDI), abductor pollicis brevis (APB), and abductor digiti minimi (ADM) muscles in each hand were created. Map features including area (primary), volume, and center of gravity were analyzed across different excitability regions (R100%, R75%, R50%, R25%). Correlations between map metrics and validated tests of hand motor function (Purdue Pegboard Test as primary) were explored. RESULTS: Twenty-four right-handed participants (range 12-18 years, median 15.5 years, 52% female) completed bilateral mapping and motor assessments with no serious adverse events or dropouts. Gender and age were associated with hand function and motor map characteristics. Full motor maps (R100%) for FDI did not correlate with motor function in either hand. Smaller excitability subset regions demonstrated reduced variance and dose-dependent correlations between primary map variables and motor function in the dominant hemisphere. CONCLUSIONS: Hand function in TDA correlates with smaller subset excitability regions of robotic TMS motor map outcomes. Refined motor maps may have less variance and greater potential to quantify interventional neuroplasticity. Robotic TMS mapping is safe and feasible in adolescents.

A Case of Neuromuscular Electrical Stimulation for Childhood Stroke Hyperkinesis: A Brief Report.

AIM: Some conditions within specific populations are so rare rigorous evidence is unavailable. Childhood hyperkinesis is one example, yet presents an opportunity to examine sensation's contribution to motor function. METHODS: The patient experienced functional difficulty from hyperkinesis as a result of childhood stroke. Home-based passive neuromuscular electrical stimulation (NMES) was implemented an hour/day, six days/week, over 6 weeks (36 hours). Clinical and robotic measures (Assisting Hand Assessment, Box and Block Test, Jebsen Taylor Test of Hand Function, Kinarm) were administered before and after the intervention and at 9 months. RESULTS: NMES was feasible and well tolerated. Clinically important gains of arm function were maintained at 9 months. Robotic measures showed improved hyperkinesis, namely reduced movement segmentation and improved target approximation, in addition to improved proprioceptive function after NMES. CONCLUSION: This case study illustrates the use of NMES within a previously unexplored population and highlights the potential importance of sensory systems to motor gains.

Effects of Transcranial Direct Current Stimulation on GABA and Glx in Children: A pilot study.

Transcranial direct current stimulation (tDCS) is a form of non-invasive brain stimulation that safely modulates brain excitability and has therapeutic potential for many conditions. Several studies have shown that anodal tDCS of the primary motor cortex (M1) facilitates motor learning and plasticity, but there is little information about the underlying mechanisms. Using magnetic resonance spectroscopy (MRS), it has been shown that tDCS can affect local levels of γ-aminobutyric acid (GABA) and Glx (a measure of glutamate and glutamine combined) in adults, both of which are known to be associated with skill acquisition and plasticity; however this has yet to be studied in children and adolescents. This study examined GABA and Glx in response to conventional anodal tDCS (a-tDCS) and high definition tDCS (HD-tDCS) targeting the M1 in a pediatric population. Twenty-four typically developing, right-handed children ages 12-18 years participated in five consecutive days of tDCS intervention (sham, a-tDCS or HD-tDCS) targeting the right M1 while training in a fine motor task (Purdue Pegboard Task) with their left hand. Glx and GABA were measured before and after the protocol (at day 5 and 6 weeks) using a PRESS and GABA-edited MEGA-PRESS MRS sequence in the sensorimotor cortices. Glx measured in the left sensorimotor cortex was higher in the HD-tDCS group compared to a-tDCS and sham at 6 weeks (p = 0.001). No changes in GABA were observed in either sensorimotor cortex at any time. These results suggest that neither a-tDCS or HD-tDCS locally affect GABA and Glx in the developing brain and therefore it may demonstrate different responses in adults.

Non-Invasive Modulation and Robotic Mapping of Motor Cortex in the Developing Brain.

Mapping the motor cortex with transcranial magnetic stimulation (TMS) has potential to interrogate motor cortex physiology and plasticity but carries unique challenges in children. Similarly, transcranial direct current stimulation (tDCS) can improve motor learning in adults but has only recently been applied to children. The use of tDCS and emerging techniques like high-definition tDCS (HD-tDCS) require special methodological considerations in the developing brain. Robotic TMS motor mapping may confer unique advantages for mapping, particularly in the developing brain. Here, we aim to provide a practical, standardized approach for two integrated methods capable of simultaneously exploring motor cortex modulation and motor maps in children. First, we describe a protocol for robotic TMS motor mapping. Individualized, MRI-navigated 12x12 grids centered on the motor cortex guide a robot to administer single-pulse TMS. Mean motor evoked potential (MEP) amplitudes per grid point are used to generate 3D motor maps of individual hand muscles with outcomes including map area, volume, and center of gravity. Tools to measure safety and tolerability of both methods are also included. Second, we describe the application of both tDCS and HD-tDCS to modulate the motor cortex and motor learning. An experimental training paradigm and sample results are described. These methods will advance the application of non-invasive brain stimulation in children.

Effects of High-Definition and Conventional Transcranial Direct-Current Stimulation on Motor Learning in Children.

Background: Transcranial direct current stimulation (tDCS) can improve motor learning in children. High-definition approaches (HD-tDCS) have not been examined in children. Objectives/Hypothesis: We hypothesized that primary motor cortex HD-tDCS would enhance motor learning but be inferior to tDCS in children. Methods: Twenty-four children were recruited for a randomized, sham-controlled, double-blinded interventional trial (NCT03193580, clinicaltrials.gov/ct2/show/NCT03193580) to receive (1) right hemisphere (contralateral) primary motor cortex (M1) 1 mA anodal conventional 1 × 1 tDCS (tDCS), (2) right M1 1 mA anodal 4 × 1 HD-tDCS (HD-tDCS), or (3) sham. Over five consecutive days, participants trained their left hand using the Purdue Pegboard Test (PPTL). The Jebsen-Taylor Test, Serial Reaction Time Task, and right hand and bimanual PPT were also tested at baseline, post-training, and 6-week retention time (RT). Results: Both the tDCS and HD-tDCS groups demonstrated enhanced motor learning compared to sham with effects maintained at 6 weeks. Effect sizes were moderate-to-large for tDCS and HD-tDCS groups at the end of day 4 (Cohen's d tDCS = 0.960, HD-tDCS = 0.766) and day 5 (tDCS = 0.655, HD-tDCS = 0.851). Enhanced motor learning effects were also seen in the untrained hand. HD-tDCS was well tolerated and safe with no adverse effects. Conclusion: HD-tDCS and tDCS can enhance motor learning in children. Further exploration is indicated to advance rehabilitation therapies for children with motor disabilities such as cerebral palsy. Clinical Trial Registration: clinicaltrials.gov, identifier NCT03193580.

Sensorimotor Robotic Measures of tDCS- and HD-tDCS-Enhanced Motor Learning in Children.

Transcranial direct-current stimulation (tDCS) enhances motor learning in adults. We have demonstrated that anodal tDCS and high-definition (HD) tDCS of the motor cortex can enhance motor skill acquisition in children, but behavioral mechanisms remain unknown. Robotics can objectively quantify complex sensorimotor functions to better understand mechanisms of motor learning. We aimed to characterize changes in sensorimotor function induced by tDCS and HD-tDCS paired motor learning in children within an interventional trial. Healthy, right-handed children (12-18 y) were randomized to anodal tDCS, HD-tDCS, or sham targeting the right primary motor cortex during left-hand Purdue pegboard test (PPT) training over five consecutive days. A KINARM robotic protocol quantifying proprioception, kinesthesia, visually guided reaching, and an object hit task was completed at baseline, posttraining, and six weeks later. Effects of the treatment group and training on changes in sensorimotor parameters were explored. Twenty-four children (median 15.5 years, 52% female) completed all measures. Compared to sham, both tDCS and HD-tDCS demonstrated enhanced motor learning with medium effect sizes. At baseline, multiple KINARM measures correlated with PPT performance. Following training, visually guided reaching in all groups was faster and required less corrective movements in the trained arm (H(2) = 9.250, p = 0.010). Aspects of kinesthesia including initial direction error improved across groups with sustained effects at follow-up (H(2) = 9.000, p = 0.011). No changes with training or stimulation were observed for position sense. For the object hit task, the HD-tDCS group moved more quickly with the right hand compared to sham at posttraining (χ 2(2) = 6.255, p = 0.044). Robotics can quantify complex sensorimotor function within neuromodulator motor learning trials in children. Correlations with PPT performance suggest that KINARM metrics can assess motor learning effects. Understanding how tDCS and HD-tDCS enhance motor learning may be improved with robotic outcomes though specific mechanisms remain to be defined. Exploring mechanisms of neuromodulation may advance therapeutic approaches in children with cerebral palsy and other disabilities.

Clinical Characteristics, Risk Factors, and Outcomes Associated With Neonatal Hemorrhagic Stroke: A Population-Based Case-Control Study.

Importance: Hemorrhage into the brain of term newborns often results in major injury and lifelong disability. The clinical epidemiology of neonatal hemorrhagic stroke (NHS) remains undefined, hindering the development of strategies to improve outcomes. Objective: To characterize the incidence, types, presentations, associated factors, and outcomes of neonatal hemorrhagic stroke. Design, Setting, and Participants: Population-based, nested case-control study. The Alberta Perinatal Stroke Project, a provincial registry, ascertained NHS cases using exhaustive diagnostic code searching (1992-2010, >2500 medical record reviews). Prospective cases were captured through the Calgary Pediatric Stroke Program (2007-2014). Participants included term neonates with magnetic resonance imaging-confirmed NHS including primary and secondary intracerebral hemorrhage, hemorrhagic transformation of ischemic injury, and presumed perinatal hemorrhagic stroke. Control infants with common data were recruited from a population-based study (4 to 1 ratio). Main Outcomes and Measures: Infants with NHS underwent structured medical record review using data-capture forms and blinded scoring of neuroimaging. Clinical risk factor common data elements were explored using logistic regression. Provincial live births were obtained from Statistics Canada. Outcomes were extrapolated to the Pediatric Stroke Outcome Measure. Results: We identified 86 cases: 51 infants (59%) with NHS, of which 32 (67%) were idiopathic, 30 (35%) were hemorrhagic transformation of primary ischemic injuries (14 with neonatal cerebral sinovenous thrombosis, 11 with hypoxic ischemic encephalopathy, and 5 with neonatal arterial ischemic stroke), and 5 were presumed perinatal hemorrhagic stroke. Sixty-two percent were male. Incidence of pure NHS was 1 in 9500 live births and 1 in 6300 for all forms. Most presented in the first week of life with seizures and encephalopathy. Acute neurosurgical intervention was rare (3 of 86 total cases; 3.5%). Temporal lobe was the most common NHS location (16 of 51 pure NHS cases; 31%). A primary cause was evident in 19 of the 51 cases of non-hemorrhagic transformation NHS (37%). Idiopathic NHS was independently associated with lower maternal age (odds ratio [OR], 0.87; 95% CI, 0.78-0.94), primiparity (OR, 2.98; 95% CI, 1.18-7.50), prior spontaneous abortion (OR, 0.11; 95% CI, 0.02-0.53), difficult fetal transition (bradycardia [OR, 15.0; 95% CI, 2.19-101.9] and low Apgar [OR, 14.3; 95% CI, 2.77-73.5]), and small for gestational age (OR, 14.3; 95% CI, 1.62-126.1). Follow-up of 50 cases at a median of 37 months demonstrated poor neurological outcomes in 21 patients (44%). Conclusions and Relevance: Neonatal hemorrhagic stroke is more common than previously reported, occurring in at least 1 in 6300 live births. Etiologies are approximately equally distributed between idiopathic, secondary, and hemorrhagic transformation. Clinical associations do not suggest a common mechanism or predictability of NHS. Recurrence is rare. Outcomes are often poor, mandating attention to prevention and rehabilitation.