Not to Cry Wolf: Distantly Supervised Multitask Learning in Critical Care

Abstract

Patients in the intensive care unit (ICU) require constant and close supervision. To assist clinical staff in this task, hospitals use monitoring systems that trigger audiovisual alarms if their algorithms indicate that a patient's condition may be worsening. However, current monitoring systems are extremely sensitive to movement artefacts and technical errors. As a result, they typically trigger hundreds to thousands of false alarms per patient per day - drowning the important alarms in noise and adding to the exhaustion of clinical staff. In this setting, data is abundantly available, but obtaining trustworthy annotations by experts is laborious and expensive. We frame the problem of false alarm reduction from multivariate time series as a machine-learning task and address it with a novel multitask network architecture that utilises distant supervision through multiple related auxiliary tasks in order to reduce the number of expensive labels required for training. We show that our approach leads to significant improvements over several state-of-the-art baselines on real-world ICU data and provide new insights on the importance of task selection and architectural choices in distantly supervised multitask learning.

References in corpus (12)

• Auto-Encoding Variational Bayes
• Improved Techniques for Training GANs
• Training Very Deep Networks
• Learning to Diagnose with LSTM Recurrent Neural Networks
• Massively Multitask Networks for Drug Discovery
• Unsupervised and Semi-supervised Learning with Categorical Generative Adversarial Networks
• Recurrent Neural Networks for Multivariate Time Series with Missing Values
• Distributed and parallel time series feature extraction for industrial big data applications
• Good Semi-supervised Learning that Requires a Bad GAN
• Triple Generative Adversarial Nets
• Distral: Robust Multitask Reinforcement Learning
• Sparse Multi-Output Gaussian Processes for Medical Time Series Prediction

Cited by in corpus (5)

• Multispectral Video Fusion for Non-contact Monitoring of Respiratory Rate and Apnea
• Perfect Match: A Simple Method for Learning Representations For Counterfactual Inference With Neural Networks
• Granger-causal Attentive Mixtures of Experts: Learning Important Features with Neural Networks
• SepAl: Sepsis Alerts On Low Power Wearables With Digital Biomarkers and On-Device Tiny Machine Learning
• AI Assisted Annotator using Reinforcement Learning