Deep Computational Phenotyping

[agitter]

http://doi.org/10.1145/2783258.2783365

Google Scholar shows several other papers that cite this one and look fairly relevant.

We apply deep learning to the problem of discovery and detection of characteristic patterns of physiology in clinical time series data. We propose two novel modifications to standard neural net training that address challenges and exploit properties that are peculiar, if not exclusive, to medical data. First, we examine a general framework for using prior knowledge to regularize parameters in the topmost layers. This framework can leverage priors of any form, ranging from formal ontologies (e.g., ICD9 codes) to data-derived similarity. Second, we describe a scalable procedure for training a collection of neural networks of different sizes but with partially shared architectures. Both of these innovations are well-suited to medical applications, where available data are not yet Internet scale and have many sparse outputs (e.g., rare diagnoses) but which have exploitable structure (e.g., temporal order and relationships between labels). However, both techniques are sufficiently general to be applied to other problems and domains. We demonstrate the empirical efficacy of both techniques on two real-world hospital data sets and show that the resulting neural nets learn interpretable and clinically relevant features.

[cgreene]

I'm going to assign this to @brettbj since it overlaps his thesis. From my first quick read - incremental training & use of a prior to push down the number of required samples are the main contributions. This also focuses on longitudinal phenotype data (like #78).