Deep Sequential Models for Suicidal Ideation From Multiple Source Data

This paper presents a novel method for predicting suicidal ideation from electronic health records (EHR) and ecological momentary assessment (EMA) data using deep sequential models. Both EHR longitudinal data and EMA question forms are defined by asynchronous, variable length, randomly sampled data sequences. In our method, we model each of them with a recurrent neural network, and both sequences are aligned by concatenating the hidden state of each of them using temporal marks. Furthermore, we incorporate attention schemes to improve performance in long sequences and time-independent pre-trained schemes to cope with very short sequences. Using a database of 1023 patients, our experimental results show that the addition of EMA records boosts the system recall to predict the suicidal ideation diagnosis from 48. 13% obtained exclusively from EHR-based state-of-the-art methods to 67. 78%. Additionally, our method provides interpretability through the t-distributed stochastic neighbor embedding (t-SNE) representation of the latent space. Furthermore, the most relevant input features are identified and interpreted medically.

Authors

• Ignacio Peis
• Pablo M. Olmos
• Constanza Vera-Varela
• Maria Luisa Barrigon
• Philippe Courtet
• Enrique Baca-Garcia
• Antonio Artes-Rodriguez

Keywords

• Predictive models
• Data models
• Recurrent neural networks
• Psychiatry
• Informatics
• Biological system modeling
• Databases
• Suicidal Ideation
• Longitudinal Study
• Neural Network
• Electronic Health Records
• Short Sequences
• Input Features
• Recurrent Neural Network
• Latent Space
• T-distributed Stochastic Neighbor Embedding
• Ecological Momentary Assessment
• Attentional Strategies
• Ecological Momentary Assessment Data
• Mental Health
• Patient Data
• Dimensionality Reduction
• Attention Mechanism
• Dense Layer
• Feed-forward Network
• Suicidal Behavior
• Temporal Correlation
• Electronic Health Record Data
• Static Information
• Neural Attention
• Global Assessment Of Functioning
• Gated Recurrent Unit
• Pre-trained Neural Network
• Attention Model
• Suicidal Thoughts
• Recall Performance
• Maximum Sequence Length
• Deep learning
• RNN
• attention
• EMA
• suicide
• Adult
• Female
• Humans
• Male
• Middle Aged
• Models, Psychological
• Neural Networks, Computer