Collaborative learning of common latent representations in routinely collected multivariate ICU physiological signals
This work addresses patient phenotyping for ICU clinicians, offering an incremental improvement by applying collaborative filtering to physiological signals.
The paper tackled patient phenotyping in ICU using multivariate physiological time series, proposing an algorithm that integrates LSTM with collaborative filtering, achieving an AUC of 0.889 and AP of 0.725 for intracranial hypertension detection.
In Intensive Care Units (ICU), the abundance of multivariate time series presents an opportunity for machine learning (ML) to enhance patient phenotyping. In contrast to previous research focused on electronic health records (EHR), here we propose an ML approach for phenotyping using routinely collected physiological time series data. Our new algorithm integrates Long Short-Term Memory (LSTM) networks with collaborative filtering concepts to identify common physiological states across patients. Tested on real-world ICU clinical data for intracranial hypertension (IH) detection in patients with brain injury, our method achieved an area under the curve (AUC) of 0.889 and average precision (AP) of 0.725. Moreover, our algorithm outperforms autoencoders in learning more structured latent representations of the physiological signals. These findings highlight the promise of our methodology for patient phenotyping, leveraging routinely collected multivariate time series to improve clinical care practices.