<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=1717549828521399&amp;ev=PageView&amp;noscript=1">

«  View All Posts

Blog Feature

By: Andrea Ramberg, RPSGT, CCSH on February 4th, 2021

Print/Save as PDF

Collaboration between Technologist and Technology: How AI Sleep Scoring Solutions are Aiding Frontline Clinicians

AI | sleep scoring | frontline clinicians

Artificial intelligence is impacting the future of virtually every industry and every human being. From freight/logistics to banking, financial services and insurance (BFSI), to cybersecurity and even marketing, many industries will be impacted by AI, with healthcare at the forefront. Per HealthTech Magazine, “There are numerous applications of AI on the market today...that can improve patient care and potentially save lives.” One such solution in the healthcare space is in the world of sleep study analysis.

Today, an AI scoring solution can perform the arduous task of scoring polysomnography (PSG) studies. Scoring a PSG takes a certified professional nearly an hour, and there can be a lot of variance between scorers, especially those trained at different locations. The goal is for an AI solution to score that same study, but in a fraction of the time, with the same levels of clinician agreement. AI is poised to completely change how sleep centers around the world operate, allowing clinicians to reinvest time back into improving the patient experience and increasing patient access and compliance.

What is AI?

Oftentimes, when one thinks of “AI,” futuristic movies like Terminator, The Matrix, I Robot, Her, Ex Machina, Wall-E and others come to mind, where an AI “Robot” either takes over the world, changes the lives of the people around them or simply creates a connection with a human being. In popular culture, AI is portrayed in the best possible light (Chappie, Wall-E) and as the biggest villain (I Robot, Terminator). In the current reality, AI is typically more of a behind-the-scenes player, rather than the star of the show.

In most cases, artificial intelligence allows for the analysis of “big data.” At its most basic concept, AI involves the development of computer algorithms, programs and software systems that can perform tasks normally completed by a human. Machine Learning (ML) is a type of AI that learns and adapts to improve its performance over time. As more data is fed through the algorithm, ML systems gain additional experience. Unlike the old auto scoring solutions, which rely on inputs from the programmer, ML systems evolve and iterate on their own. New data is the key, as a computer processes differently than humans, storing and iterating on each decision forever. Fresh batches of data ― from varying sources especially ― help the algorithm poke holes in its own logic, expanding its decision-making tree one execution at a time.

In the world of sleep scoring, there are a variety of mathematical solutions to the arduous process. Some rely on simple “if this, then that” decision-making trees, which often led to inconsistencies and repeated errors during the scoring process. As technology advances, and as new AI methods are validated and published, however, modern ML solutions powered by algorithms can soon outperform their human counterparts when it comes to scoring speed and consistency.

What Counts as AI?

AI can be as simple as a digital opponent in a game, like Deep Blue, the first AI to beat a human in Chess, or AlphaGo, which accomplished the same feat in Go. In the case of the AlphaGo AI, developers used a Monte Carlo tree search algorithm in combination with deep neural network technology. This required the team to program in the moves from previous winning events, then train the algorithm against both human and computer opponents. Over time, AlphaGo became the top “player” in the world, all because it was able to iterate and store memories and movesets beyond the capabilities of the grand masters, largely because the quantity of movesets is massive in Go compared to Chess.

When you look at the applications AI might have in healthcare, the possibilities are endless, but not without challenges. In healthcare, a strong AI solution must combine clinical, environmental and laboratory-based objective measures to allow a deeper understanding of medical disorders, all while helping to reduce the cost of care delivery and improve patient access.

To learn more about AI applications in sleep technology, view the full article in the 2020 Q4 issue of A2Zzz.

Read the A2Zzz Q4 Issue