Unmet Need
Sleep is usually measured in a clinic utilizing a technique called Polysomnography (PSG). PSG’s efficacy is largely dependent on a trained technician to accurately place EEG/EOG/EMG sensors across the patient’s head at a sleep clinic and then accurately interrupt the signals by eye. The sleep clinic setting also introduces a suboptimal testing environment: in addition to having foreign items on the patient’s head, their sleep is disturbed by being in a new environment instead of their usual home. The technician’s process of interpreting the PSG signals and converting to a hypnogram, which is used for diagnosis by a physician, is labor intensive (20-30 minutes) and inherently subjective. The highest clinical gold standard for PSG is an 85% accuracy, being limited by subjectivity assuming the sensors are placed correctly. Other items, such as Fitbits, attempt to measure sleep in an individual’s home, but have had limited efficacy, with the best only achieving 50% accuracy.
In summary, the current gold standard of PSG introduces subjectivity constraints, labor intensive, requires trained technicians to administer the test, and requires the patient to sleep in a new environment. Therefore, there is a strong need for a method that can generate a hypnogram for physicians without these limitations to better diagnose sleep conditions.
Technology Overview
Inventors at Johns Hopkins University were able to create a clinical grade in-home automatic human sleep measurement system. A novel sensor array is affixed onto the patient’s forehead within their natural home sleep environment. The sleep data is processed using artificial intelligence that is able to automatically classify sleep in accordance with current medical standards. The efficacy of this novel approach was found to match the same standard of clinical PSG (80-85% accuracy) as that performed by a trained expert. This technology is further capable of generating a hypnogram for physicians without requiring a trained technician or foreign sleep environment while removing the subjectivity errors and labor-intensive evaluations that are present with PSG.
Stage of Development
Experimental data is available.
Publication
Paper given at 2021 10th International IEEE/EMBS Conference on Neural Engineering
May 4-6, 2021, available at https://youtu.be/JOZ3aKUAmbk
