Patient Subtyping via Time-Aware LSTM Networks

By Inci M. Baytas, Cao Xiao, Xi Sheryl Zhang, Fei Wang, Anil K. Jain, Jiayu Zhou. 2017

702 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Health Information Management

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Cao Xiao
2. Jimeng Sun
3. Pascal Poupart
4. Rishab Goel
5. Fei Wang
6. Edward Choi
7. Jure Leskovec
8. Srijan Kumar

In Health Information Management

1. Cao Xiao
2. Jimeng Sun
3. Edward Choi
4. Slobodan Vucetic
5. Tian Bai
6. Shanshan Zhang
7. Brian L. Egleston
8. Fei Wang

Citations in the corpus

• Interpretable Representation Learning for Healthcare via Capturing Disease Progression through Time (2018, background): “[2] first decomposes memory cell in LSTM into long-termmemory and short-termmemory, then applies time decay factor to discount the short term memory, and finally calculates the new memory by combining the long-term memory and a discounted short-term memory.”
• Suicide Ideation Detection via Social and Temporal User Representations using Hyperbolic Learning (2021, methodology): “Uses T-LSTM (Baytas et al., 2017) which applies a monotonically decreasing function of elapsed time to weight historical tweets and utilizes BERT fine-tuned on Plutchik-based emotions for the tweet to be assessed.”
• Predicting Dynamic Embedding Trajectory in Temporal Interaction Networks (2019, positive): “as static embeddings of all users and items, as advised in TimeLSTM [52] and TimeAware-LSTM [7].”
• Time2Vec: Learning a Vector Representation of Time (2019, methodology): “There is also a set of techniques that consider time as yet another feature and concatenate time (or some hand designed features of time such as log and/or inverse of delta time) with the input [10, 33, 14, 3, 29, 55, 28, 38].”
• GRU-ODE-Bayes: Continuous modeling of sporadically-observed time series (2019, neutral): “T-LSTM (Baytas et al., 2017).”
• Med-BERT: pretrained contextualized embeddings on large-scale structured electronic health records for disease prediction (2020, methodology): “For performance measurement, we used AUC as our primary evaluation metric, which has been widely adopted by many previous studies of disease prediction [12,14,49].”
• Identify Susceptible Locations in Medical Records via Adversarial Attacks on Deep Predictive Models (2018, background): “Recurrent neural networks (RNN) are recently adopted in many predictive modeling studies and achieved state-of-the-art predictive performance, because (1) RNN can naturally handle input data of different lengths and (2) some variants such as long short-term memory (LSTM) networks have demonstrated e”
• Deep learning in mental health outcome research: a scoping review (2020, background): “For instance, Baytas et al. 122 developed a variation of LSTM-autoencoder on patient EHRs and grouped Parkinson ’ s disease patients into meaningful subtypes.”
• Deep Learning on Graphs: A Survey (2018, neutral): “To capture the temporal information of dynamic graphs, dynamic graph neural network (DGNN) [27] was proposed that used a time-aware LSTM [39] to learn node representations.”
• MetaPred: Meta-Learning for Clinical Risk Prediction with Limited Patient Electronic Health Records (2019, background): “Since MetaPred is model-agnostic, the gradient updating scheme can be easily extended to more sophisticated neural networks including various attention mechanisms or gated networks with prior medical knowledge [3, 7].”
• Opportunities and challenges in developing deep learning models using electronic health records data: a systematic review (2018, methodology): “borrowed the idea of dynamic time warping, an algorithm measuring similarity between two varying speed temporal sequences, and modeled it into the gate parameters of 2D-GRU, thus aligning EHR sequences pairwise.(11) proposed to learn a subspace decomposition of the LSTM memory, thus discounting the ”