A Novel Personalized Random Forest Algorithm for Clinical Outcome Prediction.

Machine learning algorithms that derive predictive models are useful in predicting patient outcomes under uncertainty. These are often "population" algorithms which optimize a static model to predict well on average for individuals in the population; however, population models may predict poorly for individuals that differ from the average. Personalized machine learning algorithms seek to optimize predictive performance for every patient by tailoring a patient-specific model to each individual. Ensembles of decision trees often outperform single decision tree models, but ensembles of personalized models like decision paths have received little investigation. We present a novel personalized ensemble, called Lazy Random Forest (LazyRF), which consists of bagged randomized decision paths optimized for the individual for whom a prediction will be made. LazyRF outperformed single and bagged decision paths and demonstrated comparable predictive performance to a population random forest method in terms of discrimination on clinical and genomic data while also producing simpler models than the population random forest.

Patient-Specific Modeling with Personalized Decision Paths.

Predictive models can be useful in predicting patient outcomes under uncertainty. Many algorithms employ "population" methods, which optimize a single model to perform well on average over an entire population, but the model may perform poorly on some patients. Personalized methods optimize predictive performance for each patient by tailoring the model to the individual. We present a new personalized method based on decision trees: the Personalized Decision Path using a Bayesian score (PDP-Bay). Performance on eight synthetic, genomic, and clinical datasets was compared to that of decision trees and a previously described personalized decision path method in terms of area under the ROC curve (AUC) and expected calibration error (ECE). Model complexity was measured by average path length. The PDP-Bay model outperformed the decision tree in terms of both AUC and ECE. The results support the conclusion that personalization may achieve better predictive performance and produce simpler models than population approaches.

Abdominal Pain, the Adolescent and Altered Brain Structure and Function.

Irritable bowel syndrome (IBS) is a functional gastrointestinal (GI) disorder of unknown etiology. Although relatively common in children, how this condition affects brain structure and function in a pediatric population remains unclear. Here, we investigate brain changes in adolescents with IBS and healthy controls. Imaging was performed with a Siemens 3 Tesla Trio Tim MRI scanner equipped with a 32-channel head coil. A high-resolution T1-weighted anatomical scan was acquired followed by a T2-weighted functional scan. We used a surface-based morphometric approach along with a seed-based resting-state functional connectivity (RS-FC) analysis to determine if groups differed in cortical thickness and whether areas showing structural differences also showed abnormal RS-FC patterns. Patients completed the Abdominal Pain Index and the GI Module of the Pediatric Quality of Life Inventory to assess abdominal pain severity and impact of GI symptoms on health-related quality of life (HRQOL). Disease duration and pain intensity were also assessed. Pediatric IBS patients, relative to controls, showed cortical thickening in the posterior cingulate (PCC), whereas cortical thinning in posterior parietal and prefrontal areas were found, including the dorsolateral prefrontal cortex (DLPFC). In patients, abdominal pain severity was related to cortical thickening in the intra-abdominal area of the primary somatosensory cortex (SI), whereas HRQOL was associated with insular cortical thinning. Disease severity measures correlated with cortical thickness in bilateral DLPFC and orbitofrontal cortex. Patients also showed reduced anti-correlations between PCC and DLPFC compared to controls, a finding that may reflect aberrant connectivity between default mode and cognitive control networks. We are the first to demonstrate concomitant structural and functional brain changes associated with abdominal pain severity, HRQOL related to GI-specific symptoms, and disease-specific measures in adolescents with IBS. It is possible such changes will be responsive to therapeutic intervention and may be useful as potential markers of disease progression or reversal.

The migraine brain in transition: girls vs boys.

The prevalence of migraine has an exponential trajectory that is most obvious in young females between puberty and early adulthood. Adult females are affected twice as much as males. During development, hormonal changes may act on predetermined brain circuits, increasing the probability of migraine. However, little is known about the pediatric migraine brain and migraine evolution. Using magnetic resonance imaging, we evaluated 28 children with migraine (14 females and 14 males) and 28 sex-matched healthy controls to determine differences in brain structure and function between (1) females and males with migraine and (2) females and males with migraine during earlier (10-11 years) vs later (14-16 years) developmental stages compared with matched healthy controls. Compared with males, females had more gray matter in the primary somatosensory cortex (S1), supplementary motor area, precuneus, basal ganglia, and amygdala, as well as greater precuneus resting state functional connectivity to the thalamus, amygdala, and basal ganglia and greater amygdala resting state functional connectivity to the thalamus, anterior midcingulate cortex, and supplementary motor area. Moreover, older females with migraine had more gray matter in the S1, amygdala, and caudate compared to older males with migraine and matched healthy controls. This is the first study showing sex and developmental differences in pediatric migraineurs in brain regions associated with sensory, motor, and affective functions, providing insight into the neural mechanisms underlying distinct migraine sex phenotypes and their evolution that could result in important clinical implications increasing treatment effectiveness.

Pediatric migraine prescription patterns at a large academic hospital.

BACKGROUND: Here we report the prescription patterns by drug type, age, and sex of patients at a large academic pediatric hospital. Because there are few guidelines based on outcome studies in pediatric migraine, physician treatment approaches in children vary. METHODS: Using the i2b2 query tool, we determined that over an approximately 4 year period, 4839 patients between the ages of 2 and 17 years were observed at Boston Children's Hospital for migraine with or without aura, 59% women and 41% men. RESULTS: The most common medications prescribed to this population were sumatriptan, amitriptyline, topiramate, ondansetron, and cyproheptadine. CONCLUSIONS: Our findings support recent data regarding choices of medication in the pediatric population and additionally support current studies and future investigation into controlled trials in the pediatric population.

Altered hypothalamic functional connectivity with autonomic circuits and the locus coeruleus in migraine.

The hypothalamus has been implicated in migraine based on the manifestation of autonomic symptoms with the disease, as well as neuroimaging evidence of hypothalamic activation during attacks. Our objective was to determine functional connectivity (FC) changes between the hypothalamus and the rest of the brain in migraine patients vs. control subjects. This study uses fMRI (functional magnetic resonance imaging) to acquire resting state scans in 12 interictal migraine patients and 12 healthy matched controls. Hypothalamic connectivity seeds were anatomically defined based on high-resolution structural scans, and FC was assessed in the resting state scans. Migraine patients had increased hypothalamic FC with a number of brain regions involved in regulation of autonomic functions, including the locus coeruleus, caudate, parahippocampal gyrus, cerebellum, and the temporal pole. Stronger functional connections between the hypothalamus and brain areas that regulate sympathetic and parasympathetic functions may explain some of the hypothalamic-mediated autonomic symptoms that accompany or precede migraine attacks.