The State University of New York at Buffalo and Sleep Solutions, Inc. have joined together to fight two forms of life-threatening sleeping disorders which impact millions. In Obstructive Sleep Apnea (OSA), the airway relaxes and closes causing patients to stop breathing at least five times an hour. Each non-breathing period is called an apnea. Over 18 million Americans suffer from OSA, making it as prevalent as diabetes or asthma.

In the second form, Cheyne-Stokes Respiration (CSR-CSA), the sleeping patient has cycles of slowing breath, then no breathing, followed by increas­ingly rapid breathing. This breathing pattern is caused by the body’s inability to regulate CO2 levels adequately. CSR affects 4.6 million Americans.

Besides startling spouses, these sleeping disorders have serious consequences. About one-third of hypertension patients have OSA and one half of congestive heart failure cases have either OSA or CSR-CSA. By disrupting sleeping patterns, altering the blood oxygen content and causing biochemical alterations, OSA and CSR-CSA actually aggravate coronary conditions increasing the mortality rates by 50%. OSA is also linked to diabetes, depression, epilepsy and headaches.

In addition, the drowsiness and irritabil­ity caused by sleep deprivation leads to family stress, personality changes and a six fold increase in vehicular death. All told, undiagnosed OSA patients consume twice as many healthcare resources and spend three times longer in the hospital as compared to patients without OSA. The combination of high prevalence and serious health risks makes obstructive sleep apnea as large a public health hazard as cigarette smoking.

Despite the serious nature of these conditions, less than 5% of the afflicted are diagnosed. This is partly due to the diagnostic procedure, which requires an expensive and inconvenient overnight clinic stay. During the stay, the patient sleeps while attached to dozens of chest and finger sensors and is monitored by a camera. At the conclusion of the exam, hours of data are manually inspected by the physician. Frequently, the stress of unfamiliar surroundings alters the patient’s natural sleep patterns, masking the symptoms of OSA. This can make extra nights of observation necessary. In addition, there is a severe shortage of sleep disorder diagnostic centers and sleep technicians (polysomnographers). OSA and CSR-CSA are treatable disorders, making the low diagnostic rate particularly tragic.

Companies such as Sleep Solutions, Inc. have worked hard to simplify sleep disorder diagnosis by allowing it to be performed at home. First, pulse oximetry identification of OSA sufferers has been refined. Pulse oximetry works on a simple principle. Most everyone is familiar with covering a flashlight with their finger and making it “glow” red as if they were ET. Pulse oximetry works the same way but quantifies the red color as a reflection of the oxygen content of the blood in the finger. The oxygen content of the blood decreases if the sleeper is not breathing during apnea. Second, pulse oximetry devices have physically shrunk in size, making them portable. The units can now be placed in the patient’s home, saving money, allowing patients to sleep in the comfort of their own beds and providing more accurate diagnostic readings.

UB inventors Ali A. El-Solh, M.D., Brydon J. B. Grant, M.D. and Jacek Dmochowski (now with the University of North Carolina at Charlotte) have now supplied another advancement for at-home diagnosis. They created a unique artificial intelligence software to use artificial neural networks to analyze hours and hours of pulse oximetry data. Figure 1. A simple Artificial Neural Network is shown above. These computer programs were originally designed to mimic the human brain at a simple level. Each box represents a processing element or “unit” analogous to a neuron. The input units process information they receive and communicate the result to a hidden layer or layers of units. Each of the results sent to the hidden layer is weighted. After processing the results from the input layer, the hidden layer communicates the new results to the output layer. The output layer then compares the generated results against the experimental observation and uses a back-propagation algorithm to adjust the weights assigned to the various inputs in order to optimize the final result. In this way, the program can “learn” from its experiences.

Artificial neural networks (ANN), as the name implies, mimic information processing as done by the brain (see fig­ure 1). ANN programs are trainable and are very good at sophisticated pattern recognition. The input data consists of several measurements taken from the pulse oximetry readings. Previous to this software, these measurements were analyzed separately to make a diagnosis. This software is able to combine these measurements to create a much more accurate diagnosis than previously possible. Further, it is able to separately diagnosis OSA and CSR-CSA based solely on pulse oximetry data. Previously, pulse oximetry was intended to screen only for OSA.

“Sleep Apnea is an important public health problem because it is common and excessive daytime sleepiness can result in industrial and road traffic accidents” stated Dr. Grant. “We became involved because it is a disease that even expert physicians have difficulty in making the correct diagnosis. Over the last decade, there have blossomed a number of new statistical methods that have been designed to enhance predic­tive properties. We have been exploring them in the hope of helping our colleagues to make this diagnosis more accurately”.

In March, the University at Buffalo signed a licensing agreement with Sleep Solutions, Inc. (SSI). Sleep Solutions is a privately owned company based in Maryland. Their strong market presence and product line in area of sleep apnea diagnosis as well as clinical-trial experience make them an ideal partner to commercialize this technology. According to Michael J. Thomas, President and CEO of Sleep Solutions, “This technology represents a substan­tive advancement in the way OSA can be diagnosed. This new product will broaden our portfolio of services of less expensive, more-patient-friendly diagnostic testing products delivered directly to OSA patients in their homes…SSI has contracted with many of the leading managed care organizations throughout the U.S. and developed strong brand awareness of innovation, cost-effectiveness and high-quality services with our core product, NovaSom™ QSG™.”

The average sleep lab receives $1,200 per sleep study; SSI’s goal is to continue to drive down the cost of diagnostic testing for OSA and CSR by at least 50% or more with these new technolo­gies. SSI intends to apply to the FDA for a 510k approval to market the University at Buffalo technology. If the application process is successful, SSI will be able to market the new technology to managed care organizations in early 2006.

With its NovaSom™ QSG™ At-Home Diagnostic System, Sleep Solutions(tm), Inc. offers unassisted, at-home patient testing for obstructive sleep apnea (OSA) through its Internet-Enabled NovaSom™ QSG™ Service. When ordered by a physician, the NovaSom™ QSG™ System is shipped directly to the patient’s home. The NovaSom™ QSG™ collects up to three nights of sleep data and then it is returned to Sleep Solutions where the data are downloaded. A comprehensive summary is sent to the physician for review and diagnosis. For more information, contact Sleep Solutions at 877-753-3776 or visit http://www.sleep-solutions.com

Contact information: University at Buffalo Office of Science, Technology Transfer and Economic Outreach (STOR): www.stor.buffalo.edu

Sleep Solutions, Inc.: www.sleep-solutions.com Michael L. Fowler, Ph.D. is a Commercialization Manager for the University at Buffalo Office of Science, Technology Transfer and Economic Outreach (STOR).