STQS: Interpretable multi-modal Spatial-Temporal-seQuential model for automatic Sleep scoring
نویسندگان
چکیده
Sleep scoring is an important step for the detection of sleep disorders and usually performed by visual analysis. Since manual time consuming, machine-learning based approaches have been proposed. Though efficient, these algorithms are black-box in nature difficult to interpret clinicians. In this paper, we propose a deep learning architecture multi-modal scoring, investigate model's decision making process, compare reasoning with annotation guidelines AASM manual. Our architecture, called STQS, uses convolutional neural networks (CNN) automatically extract spatio-temporal features from 3 modalities (EEG, EOG EMG), bidirectional long short-term memory (Bi-LSTM) sequential information, residual connections combine features. We evaluated our model on two large datasets, obtaining accuracy 85% 77% macro F1 score 79% 73% SHHS in-house dataset, respectively. further quantify contribution various architectural components conclude that adding LSTM layers improves performance over CNN, while does not. interpretability results show output well aligned guidelines, therefore, decisions correspond domain knowledge. also models single-channel suggest future research should focus improving models.
منابع مشابه
Sequential Bayesian Optimisation for Spatial-Temporal Monitoring
Determine a non-myopic solution to the sequential decision making problem of monitoring and optimising a space and time dependent function using a moving sensor. Contributions: Sequential Bayesian Optimisation (SBO) Formulate SBO as a Partially Observed Markov Decision Process (POMDP). Find non-mypic solution for the POMDP analog of SBO using MonteCarlo Tree Search (MCTS) and Upper Confidence B...
متن کاملMulti-modal Multi-task Learning for Automatic Dietary Assessment
We investigate the task of automatic dietary assessment: given meal images and descriptions uploaded by real users, our task is to automatically rate the meals and deliver advisory comments for improving users’ diets. To address this practical yet challenging problem, which is multi-modal and multi-task in nature, an end-to-end neural model is proposed. In particular, comprehensive meal represe...
متن کاملEffectiveness of sequential automatic-manual home respiratory polygraphy scoring.
Automatic home respiratory polygraphy (HRP) scoring functions can potentially confirm the diagnosis of sleep apnoea-hypopnoea syndrome (SAHS) (obviating technician scoring) in a substantial number of patients. The result would have important management and cost implications. The aim of this study was to determine the diagnostic cost-effectiveness of a sequential HRP scoring protocol (automatic ...
متن کاملAutomatic multi-modal dialogue scene indexing
An automatic algorithm for indexing dialogue scenes in multimedia content is proposed. The content is segmented into dialogue scenes using the state transitions of a hidden Markov model (HMM). Each shot is classified using both audio and visual information to determine the state/scene transitions for this model. Face detection and silence/speech/music classification are the basic tools which ar...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Artificial Intelligence in Medicine
سال: 2021
ISSN: ['1873-2860', '0933-3657']
DOI: https://doi.org/10.1016/j.artmed.2021.102038