Sleeping does not mean dreams for everybody. In fact, for an estimated 20 to 50% of the overall population, it can resemble a nightmare more than a dream due to sleep apnea. Imagine your breathing repeatedly stopping for seconds at a time during sleep without you realizing it, but your brain does. It reacts by interrupting your sleep cycle to restore normal breathing. The result? Non-restoring rest, constant daily somnolence, headaches, and potentially cardiovascular implications. And all this often without the person ever knowing why.
If a partner or roommate happens to notice your breathing pauses, and you comment on this or describe the previous symptoms to a doctor, you might be referred to a sleep specialist. Eventually, this leads (after a long wait) to an overnight stay at a sleep laboratory for a polysomnography, the current gold-standard test for diagnosing sleep apnea. Sounds like the end of the nightmare? Not yet. During the test, you will spend the night in a hospital bed, covered in wires and electrodes tracking your body’s signals. Later, a specialist manually reviews and scores hours of recordings to assess whether you truly have sleep apnea and how severe it is
The drawbacks of this process are clear: long waiting lists for access to a bed in the sleep laboratory, labor-intensive analysis for healthcare professionals, and a clinical setting so uncomfortable it may alter your natural sleep, potentially skewing the diagnosis.
Fortunately, artificial intelligence (AI) is opening new doors. By focusing on a limited number of physiological signals typically included in polysomnography (such as oxygen saturation, airflow, or electrocardiogram activity) researchers can streamline the diagnostic process. The recordings of those signals can be performed at home, using portable devices. AI algorithms then analyze the data, detecting apnea events and providing automated, fast assessments for suspected cases.
What’s more, the integration of AI with explainable artificial intelligence techniques, which offer trustworthy explanations of the decisions taken by the AI algorithms, adds a crucial layer of trust. Medical experts can interpret these algorithmic outcomes with confidence, ensuring transparency in diagnosis and opening the door for real-time, patient-centered care.
As AI continues to evolve, it may soon transform sleep medicine by making diagnosis faster, cheaper, and more accessible. For those unknowingly living with sleep apnea, this technology offers not only the promise of better sleep, but a safer and healthier life.
Postdoctoral researcher at the Biomedical Engineering Group from the University of Valladolid. His research is primarily focused on applying signal processing techniques to biomedical signals to characterize and help diagnose sleep apnea.