DeepSleep: Fast and Accurate Delineation of Sleep Arousals at Millisecond Resolution by Deep Learning

Abstract

Sleep arousals are transient periods of wakefulness punctuated into sleep. Excessive sleep arousals are associated with many negative effects including daytime sleepiness and sleep disorders. High-quality annotation of polysomnographic recordings is crucial for the diagnosis of sleep arousal disorders. Currently, sleep arousals are mainly annotated by human experts through looking at millions of data points manually, which requires considerable time and effort. Here we present a deep learning approach, DeepSleep, which ranked first in the 2018 PhysioNet Challenge for automatically segmenting sleep arousal regions based on polysomnographic recordings. DeepSleep features accurate (area under receiver operating characteristic curve of 0.93), high-resolution (5-millisecond resolution), and fast (10 seconds per sleep record) delineation of sleep arousals.

Fast decoding cell type–specific transcription factor binding landscape at single-nucleotide resolution

“…Unlike classical machine learning approaches that require complicated feature crafting and engineering [22,28,29] , neural network models automatically learn the informative features and factors contributing to TF binding, largely increasing the flexibility and scalability [30,31] . Moreover, the limited hard drive size and computer memory usually restrict the complete search of the feature space in classical machine learning approaches.…”

“…The overall AUPRC, or the gross AUPRC, is defined as U P RC (R ) A = ∑ j P j j − R j+1 P j = total number of sites with predicted probability (j/1000) or greater number of T F binding sites with predicted probability (j/1000) or greater R j = total number of T F binding sites number of T F binding sites with predicted probability (j/1000) or greater where the Precision ( and Recall ( ) were calculated at each cutoff j and j = 0, 0.001, ) P j R j 0.002, …, 0.998, 0.999, 1. When multiple chromosomes are under consideration, this overall AUPRC is similar to the "weighted AUPRC", which is different from simply averaging the AUPRC score of all chromosome [30] . This is because the overall AUPRC considers the length of each chromosome and longer chromosomes contribute more to the overall AUPRC, resulting in a more accurate evaluation of the performance.…”

Leopard: fast decoding cell type-specific transcription factor binding landscape at single-nucleotide resolution

“…The overall AUPRC, or the gross AUPRC, is defined as where the Precision ( and Recall ( ) were calculated at each cutoff j and j = 0, 0.001, ) P j R j 0.002, …, 0.998, 0.999, 1. When multiple chromosomes are under consideration, this overall AUPRC is similar to the "weighted AUPRC", which is different from simply averaging the AUPRC score of all chromosome [30] . This is because the overall AUPRC considers the length of each chromosome and longer chromosomes contribute more to the overall AUPRC, resulting in a more accurate evaluation of the performance.…”