Discovering Pediatric Asthma Phenotypes Based on Response to Controller Medication Using Machine Learning.

Discovering Pediatric Asthma Phenotypes Based on Response to Controller Medication Using Machine Learning.

Ann Am Thorac Soc. 2017 Oct 19;:

Authors: Ross MK, Yoon J, van der Schaar A, van der Schaar M

Abstract
RATIONALE: Pediatric asthma has variable underlying inflammation and symptom control. Approaches to address this heterogeneity such as clustering methods to find phenotypes and predicting outcomes have been investigated. However, clustering based upon the relationship between treatment and clinical outcome has not been performed. Also, machine learning approaches for long-term outcome prediction in pediatric asthma have not been studied in depth.
OBJECTIVES: Our objectives were to use our novel machine learning algorithm, Predictor Pursuit (PP) to discover pediatric asthma phenotypes based on treatment response to different asthma controller medications, to predict outcomes of children with asthma over time, and to identify the most indicative features of asthma control for the discovered pediatric phenotypes.
METHODS: We applied PP to the Childhood Asthma Management Program (CAMP) study data (n=1,019) to discover phenotypes based on treatment outcome (budesonide vs. nedocromil). We confirmed PP's ability to discover phenotypes using the Asthma Clinical Research Network (ACRN)/Childhood Asthma Research and Education (CARE) network data. We next predicted children's asthma outcomes over time and compared PP's performance to traditional prediction methods. Last, we determined the clinical features most correlated with asthma control in the discovered phenotypes.
RESULTS: Four phenotypes were discovered in both datasets: allergic-not-obese (A(+)/O(-)), obese-not-allergic (A(-)/O(+)), allergic-and-obese (A(+)/O(+)), and neither-obese-nor-allergic (A(-)/O(-)). Of the well-controlled children in the CAMP dataset, we found more non-obese children treated with budesonide than nedocromil (p=0.015) and more obese children treated with nedocromil than budesonide (p=0.008). Within the obese group, more A(+)/O(+) children were well-controlled with nedocromil than budesonide (p=0.022) or placebo (p=0.011). The PP algorithm performed significantly better (p<0.001) than traditional machine learning algorithms for both short and long-term asthma control prediction. In the short-term (four months), control state followed by airways hyperreactivity was most predictive of outcome regardless of treatment or phenotype. In the long-term (one year), less airways hyperreactivity and serum eosinophils were most predictive of better control regardless of phenotype or assigned medication.
CONCLUSIONS: Advanced statistical machine learning approaches can be a powerful tool to discover phenotypes based upon treatment response and to predict outcomes in complex medical conditions such as asthma.

PMID: 29048949 [PubMed - as supplied by publisher]

from # All Medicine by Alexandros G. Sfakianakis via Alexandros G.Sfakianakis on Inoreader http://ift.tt/2zpni8B
via IFTTT