Machine Learning of Patient Characteristics to Predict Admission Outcomes in the Undiagnosed Diseases Network

Authors:

Abstract
Rare and undiagnosed diseases are paradigmatic for the era of precision medicine. Although there is no unique definition for such diseases,1 a disease is typically considered rare if its prevalence is less than 1 per 1250 in the US,2 per 2000 in Europe,3 and per 2500 and 10 000 in Japan and Australia,4 respectively. Many undiagnosed diseases are likely to include a genetic component to their pathogenesis, and yet patients will find themselves on a protracted journey from one specialist to another without diagnosis even in this era of genomic sequencing.5,6 The onset in about half of undiagnosed diseases occurs at birth or during infancy, and these diseases are often associated with premature mortality and disability that is present throughout a patient’s life.1,7-9 According to the Office of Rare Disease Research at the National Institutes of Health, approximately 6% of the inquiries made to the Genetic and Rare Disease Information Center are in reference to an undiagnosed disease.10 In addition, although a large number of rare diseases exist (>7000 different types10), only a small fraction of these diseases (approximately 350 rare diseases) affect more than 80% of all patients with rare diseases.11

This publication has 16 references indexed in Scilit: