CIOTechie_ArtificialIntelligence_Sleeping_Apnea_Hypopnea_Arousal_

Artificial Intelligence Can Predict Sleep Disorders Like Sleep Apnea, Hypopnea, And Arousal

Artificial Intelligence News

Artificial_Intelligence-Can-Predict-Sleep-Disorders-Like-Sleep-Apnea-Hypopnea-And-Arousal Artificial Intelligence Can Predict Sleep Disorders Like Sleep Apnea, Hypopnea, And ArousalArtificial Intelligence has the capability to transform the businesses not only tech firms but in healthcare sector too. In the healthcare industry, it uses to accurately diagnosis patients’ medical data, collecting and distributing medical reports between doctors and patients, as well as help advancing health care technologies to provide superior service. Now, AI is able to detect restless sleep after the researchers at Stanford and Université Paris-Saclay, in April last year, proposed a system that can envisage location, duration, and type of sleep event in EEG charts. Additionally, Oxford scientists, last November, explained a framework which could automatically detect REM sleep behavior disorder.

However, a method explained in a paper entitled with “SleepNet: Automated sleep disorder detection via dense convolutional neural network” takes a slightly different spike. Rather than look for patterns of disordered sleep in slices of sensor data, it takes into account various data collected during polysomnography, sleep studies. The research team architected its Artificial Intelligence system atop a convolutional neural network, which is usually implemented to visual imagery analysis, with a remapping mechanism to shorten the network decision-making process. They, to improve generalization, utilized multitask learning mechanism to look for correlations among three conditions Arousal, Apnea, and Hypopnea. And to train that mechanism, they sourced the 12 measurement channels provided in the open source PhysioNet challenge corpus that holds manually annotated polysomnography data from 1,985 patients monitored at Massachusetts General Hospital’s sleep laboratory.

Afterward, the researchers then repeated the full training process a total of 4 times across various fourfold of training and validation data, with 794 samples per fold in the training set and 100 validations and 100 consistent testing records. Then they found an ensemble strategy, wherein one that tapped manifold trained models improved in performance, compared to several single-morel strategies. Though it sometimes overestimated the apnea-hypopnea severity, researchers claimed that it’s able to predict arousal, apnea, and hypopnea precisely.