Machine learning OCT predictors of progression from intermediate age-related macular degeneration to geographic atrophy and vision loss.

Objective: To describe optical coherence tomography (SD-OCT) features, age, gender, and systemic variables that may be used in machine/deep learning studies to identify high-risk patient subpopulations with high risk of progression to geographic atrophy (GA) and visual acuity (VA) loss in the short term. Design: prospective, longitudinal study. Subjects: We analyzed imaging data from patients with iAMD (N= 316) enrolled in Age-Related Eye Disease Study 2 (AREDS2) Ancillary SD-OCT with adequate SD-OCT imaging for repeated measures. Methods: Qualitative and quantitative multimodal variables from the database were derived at each yearly visit over 5 years. Based on statistical analyses developed in the field of cardiology, an algorithm was developed and used to select person-years without GA on colour fundus photography or SD-OCT at baseline. The analysis employed machine learning approaches to generate classification trees. Eyes were stratified as low, average, above average and high risk in 1 or 2 years, based on OCT and demographic features by the risk of GA development or decreased VA by 5+ and 10+ letters. Main outcome measures: new onset of SD-OCT-determined GA and VA loss. Results: We identified multiple retinal and subretinal SD-OCT and demographic features from the baseline visit, each of which independently conveyed low to high risk of new-onset GA or VA loss on each of the follow-up visits at 1 or 2 years. Conclusion: We propose a risk-stratified classification of iAMD based on the combination of OCT-derived retinal features, age, gender and systemic variables for progression to OCT-determined GA and/or VA loss. After external validation, the composite early endpoints may be used as exclusion or inclusion criteria for future clinical studies of iAMD focused on prevention of GA progression or VA loss.

A spatio-temporal absorbing state model for disease and syndromic surveillance.

Reliable surveillance models are an important tool in public health because they aid in mitigating disease outbreaks, identify where and when disease outbreaks occur, and predict future occurrences. Although many statistical models have been devised for surveillance purposes, none are able to simultaneously achieve the important practical goals of good sensitivity and specificity, proper use of covariate information, inclusion of spatio-temporal dynamics, and transparent support to decision-makers. In an effort to achieve these goals, this paper proposes a spatio-temporal conditional autoregressive hidden Markov model with an absorbing state. The model performs well in both a large simulation study and in an application to influenza/pneumonia fatality data.

Cherry-picking for complex data: robust structure discovery.

Complex data often arise as a superposition of data generated from several simpler models. The traditional strategy for such cases is to use mixture modelling, but it can be problematic, especially in higher dimensions. This paper considers an alternative approach, emphasizing data exploration and robustness to model misspecification. The strategy is applied to problems in regression, cluster analysis and multidimensional scaling. The approach is illustrated through simulation and the analysis of several datasets.

Functional brain maturation in neonates as measured by EEG-sleep analyses.

OBJECTIVE: Seven measures of neonatal EEG-sleep behavior were evaluated using multivariate analyses to ascertain if physiologic differences exist between healthy full- and preterm cohorts. METHODS: A total of 381 24-channel EEG-sleep studies were analyzed, including 125 recordings on 50 healthy fullterm and 256 recordings on 59 asymptomatic preterm infants between 28 and 70.6 weeks post-conceptional age. One EEG study for each subject was randomly assigned (109 studies) within the time window of 38-44 weeks post-conceptional age. A multivariate analytic procedure was applied to the data sets, by which a 'dysmaturity index' was assigned for each infant, based on 7 EEG-sleep measures. This index was defined in terms of the distance from the fullterm group's centroid (i.e. Mahalanobis distance). Receiver-operating characteristic curves (ROCs) were calculated for several different combinations of 7 EEG-sleep measures to describe differences between neonatal cohorts. RESULTS: The ROC curve corresponding to all 7 EEG-sleep measures covered the substantially largest area among the curves for the sets of variables considered, suggesting that all 7 measures of sleep behavior were required to best discriminate between cohorts. CONCLUSIONS: This methodology exemplifies how EEG-sleep analyses can be applied to the study of functional brain maturation of infants at risk for neurodevelopment problems. SIGNIFICANCE: Changes in EEG-sleep behavior in the preterm infant may represent altered activity-dependent development of neural circuitry, resulting in remodeling of the immature brain as a reflection of adaptation to conditions of prematurity.

Temperature differences during sleep between fullterm and preterm neonates at matched post-conceptional ages.

OBJECTIVE: Altered physiologic behaviors during sleep have been described for healthy preterm neonates at post-conceptional fullterm ages. These differences may reflect brain dysmaturity as a result of conditions of prematurity. The present study examines if differences in state-specific temperature changes exist in a healthy preterm cohort as another expression of brain dysmaturity. METHODS: Rectal and skin temperatures during sleep state transitions are reported in 59 asymptomatic post-conceptional age term infants, comparing 25 full term and 34 preterm infants. Three-hour 24-channel electroencephalogram (EEG)-sleep studies were recorded for each child. One of 4 sleep states were assigned for each of 7339 min, based on both cerebral and non-cerebral measures. For each study, average rectal and skin temperatures for each sleep state were calculated. Repeated measures MANOVA were performed using 4 explanatory variables, average skin and rectal temperatures and variance of rectal and skin temperatures, comparing preterm/fullterm status and 4 sleep states. RESULTS: Rectal temperature differences between neonatal cohorts during specific sleep states were noted: F=8.58, P<0.0001. Significant differences were noted for both average and variance of rectal temperatures during all 4 sleep states with higher temperatures in the preterm group. For all neonates, both skin and rectal temperature differences were also noted among sleep states (F=4.22, P<0.0004). Differences were specifically noted between two specific EEG segments, mixed frequency active sleep and tracé alternant quiet sleep (P<0.0004). CONCLUSIONS: In summary, significant differences in temperatures were noted across sleep state transitions for two neonatal cohorts, with higher average rectal temperatures in the preterm cohort. These findings highlight an altered measure of brain function during sleep in preterm infants affecting temperature regulation. This altered physiologic behavior reflects adaptation of the infant's brain function to conditions of prematurity which may contribute to vulnerabilities at older ages.

Neonatal EEG-sleep disruption mimicking hypoxic-ischemic encephalopathy after intrapartum asphyxia.

OBJECTIVES: EEG-sleep organization of asphyxiated and non-asphyxiated full-term neonates was compared during the first 3 days after birth. BACKGROUND: Aggressive fetal and neonatal resuscitative efforts have reduced the severe expression of the neonatal brain disorder termed hypoxic-ischemic encephalopathy. Neonates may alternatively express altered EEG-sleep organization over the first days of life after asphyxia which may mimic mild or moderate hypoxic-ischemic encephalopathy. None of ten asphyxiated infants had EEG-confirmed seizures or pharmacologically-induced encephalopathies. All asphyxiated infants expressed fetal distress on fetal heart monitoring prior to delivery, and required neonatal resuscitation, as reflected in depressed 1, 5, and 10 min Apgar scores. Moderate to severe metabolic acidosis was also documented at birth in the asphyxiated group. All ten asphyxiated infants displayed either hyperalertness/irritability or somnolence/lethargy during the first 24 h after birth, suggesting mild to moderate post-asphyxial encephalopathy. Twenty-two 1 h 21-channel EEG polygraphic studies were obtained from the first through third days of life on nine asphyxiated infants and scored for EEG-sleep states. Studies on 23 non-asphyxiated newborns were also obtained between 1 and 3 days of life and scored for EEG-sleep state. EEG-sleep states were assigned to every minute of each record by visual analyses, without knowledge of the presence or absence of asphyxia. Comparisons of active, quiet, and indeterminate sleep percentages between neonatal groups were performed. Nested MANOVA was used which took into account multiple observations per child in the asphyxiated group. RESULTS: The percent of active sleep was 44.7% (+/-14.7), the percent of quiet sleep was 38.7% (+/-14.3), and the percent of indeterminate sleep was 13.3% (+/-11.4) in the non-asphyxiated group. The percent of active sleep was 18.9% (+/-18.5), the percent of quiet sleep was 46.5% (+/-21.1), and the percent of indeterminate sleep was 33.4% (+/-19.7) in the asphyxiated group. A significant decrease in active sleep (F=39.5, P<0.0001), and significant increases in quiet sleep (F=4.6, P<0.05) and indeterminate sleep (F=21.5, P<0.0005) were noted in the asphyxiated group. Shorter active sleep bout lengths were noted (F=21.8, P<0.001), while the quiet sleep bout lengths remained unchanged for the asphyxiated group. CONCLUSIONS: An increased percentage of quiet sleep and indeterminate sleep at the expense of decreased active sleep reflects postnatal brain adaptation to asphyxia in infants despite the absence of overt clinical or electrographic evidence of hypoxic-ischemic encephalopathy. Brain adaptation in newborns after acute asphyxial stress may be expressed as altered sleep organization, despite clinical signs which may masquerade as mild to moderate post-asphyxial encephalopathy. EEG-sleep studies can assist in a more accurate classification of newborn encephalopathy that does not satisfy the criteria for hypoxic-ischemic encephalopathy.