Join Discord
Created at 6am, Apr 5
Ms-RAGArtificial Intelligence
0
Alzheimer’s disease detection in PSG signals
Contract ID
IHbaflc9lJmG4geAIxu5KJELX-lH_Vt5MN7HXeRYbmc
File Type
PDF
Entry Count
72
Embed. Model
jina_embeddings_v2_base_en
Index Type
hnsw

Lorena Gallego-Vinaras, Juan Miguel Mira-Tom´as, Anna Michela Gaeta, Gerard Pinol-Ripoll, Ferran Barbe,Pablo M. Olmos, Arrate Mu˜noz-Barrutia, Fellow, IEEEAbstract—Alzheimer’s disease (AD) and sleep disorders exhibit a close association, where disruptions in sleep patterns often precede the onset of Mild Cognitive Impairment (MCI) and early stage AD. This study delves into the potential of utilizing sleep related electroencephalography (EEG) signals acquired throughpolysomnography (PSG) for the early detection of AD. Our primary focus is on exploring semi-supervised Deep Learning techniques for the classification of EEG signals due to the clinical scenario characterized by the limited data availability.The methodology entails testing and comparing the performance of semi-supervised SMATE and TapNet models, benchmarked against the supervised XCM model, and unsupervised Hidden Markov Models (HMMs). The study highlights the significance of spatial and temporal analysis capabilities, conducting independent analyses of each sleep stage. Results demonstrate the effectiveness of SMATE in leveraging limited labeled data, achieving stable metrics across all sleep stages, and reaching 90% accuracy in its supervised form. Comparative analyses reveal SMATE’s superior performance over TapNet and HMM,while XCM excels in supervised scenarios with an accuracy range of 92 − 94%. These findings underscore the potential of semi-supervised models in early AD detection, particularly in overcoming the challenges associated with the scarcity of labeled data. Ablation tests affirm the critical role of spatio-temporalfeature extraction in semi-supervised predictive performance, and t-SNE visualizations validate the model’s proficiency in distinguishing AD patterns. Overall, this research contributes to the advancement of AD detection through innovative Deep Learning approaches, highlighting the crucial role of semisupervisedlearning in addressing data limitations.Index Terms—Alzheimer’s disease (AD), Deep Learning (DL), Electro Encephalogram (EEG), Mild Cognitive Impairment (MCI), polysomnography (PSG), semi-supervised models.

However, when trained with only 10% of the samples labeled, there was a significant decrease in performance. Despite this decline, it still outperformed TapNet in its supervised mode, which exhibited inconsistent curves, notably in the N3 stage. This observation suggests the robustness of the SMATE model. To assess the potential bias arising from varying age ranges among patients, tests were conducted to evaluate its impact. Three categories were defined, encompassing patients aged 30-47 years in the first group, 48-69 in the second, and 7084 in the third, each group consisting of an equal number of individuals. Following this categorization, the models were retrained to classify signals within each new age category. The tests were unsuccessful, as none of the metrics surpassed 0.4 in any sleep stage.
id: fbf2d825434075fe667978e256f11d45 - page: 6
Statistical Analysis: An ANOVA test revealed statistically significant differences among the models. The post-hoc analysis identified these discrepancies, particularly between SMATE and XCM, which have higher performance than TapNet. In the unsupervised comparison, SMATE, TapNet and HMM showed no statistically significant differences, exhibiting similar performance in all sleep phases. ROC curves were plotted for each model and sleep stage, indicating the trade-off between True Positive Rate and False Positive Rate at different thresholds (Fig. 5). In its supervised
id: deac5f3e0d13c3d7dff69009295c3c47 - page: 6
B. Ablation test The elimination of the spatial block yielded significant changes in the models capability for processing contextual 6 (a) (b) (c) Fig. 5: Comparative analysis of SMATE supervised, SMATE with 10% labeled samples and TapNet supervised model performances. This figure displays the accuracy, F1-score metrics (with their standard deviations), and ROC curves for each model across different sleep stages. Each panel, from left to right, represents the results of each fold for the corresponding model: (a) Depicts the SMATE fully supervised model, showcasing its performance consistency across sleep stages; (b) Illustratres the SMATE 10% supervised model, showcasing its variability across different stages while still outperforming (c), which represents the TapNet fully supervised model, revealing greater variability compared to SMATE across the sleep stages.
id: 184e1046dbbf068e47ea24667b8ecde9 - page: 6
In the semi-supervised models, the absence of spatial information resulted in a noticeable decrease in both consistency and coherence. Specifically, TapNet was the most affected model, exhibiting a random classification considering the ROC/AUC metrics. Similarly, SMATE demonstrated reduced metric values across all phases, yet the decline was less severe compared to TapNet, with REM sleep showing the most significant decrease. Meanwhile, the supervised XCM model showed considerable robustness against the removal of spatial information, with only a minor impact on its overall performance.
id: 8b224aa03ea65a3f62a843718e3c04f7 - page: 7
How to Retrieve?
# Search

curl -X POST "https://search.dria.co/hnsw/search" \
-H "x-api-key: <YOUR_API_KEY>" \
-H "Content-Type: application/json" \
-d '{"rerank": true, "top_n": 10, "contract_id": "IHbaflc9lJmG4geAIxu5KJELX-lH_Vt5MN7HXeRYbmc", "query": "What is alexanDRIA library?"}'
        
# Query

curl -X POST "https://search.dria.co/hnsw/query" \
-H "x-api-key: <YOUR_API_KEY>" \
-H "Content-Type: application/json" \
-d '{"vector": [0.123, 0.5236], "top_n": 10, "contract_id": "IHbaflc9lJmG4geAIxu5KJELX-lH_Vt5MN7HXeRYbmc", "level": 2}'