The percentage of chronic disease (CD) patients is climbing to reach almost one-fifth globally, posing immediate threats on healthcare expenditure (in Qatar it reaches ~2.5% of the national GDP), and rising causes of death (more than 76% of deaths in Qatar are owed to CDs). This calls for transforming healthcare systems away from one-on-one patient treatment, to improve services, access and scalability, while reducing costs.
The deployment of mobile and wearable devices in healthcare along with the advances of wireless networking and cloud-based technologies will enable mobile health (mHealth) systems to operate in conjunction with invaluable sets of data that are amassed from wide range of sources. Smart Health (sHealth) can be considered the evolution of mHealth, where collected medical and physiological data can be efficiently mined at the patient level to better understand the patient context, hence optimizing the transmission and integration of diagnostic information from healthcare professionals at the cloud level, while facilitating immediate response in cases of emergency. Such concept is set to revolutionize healthcare delivery by giving CD patients better control of their medical situation and allow the public to continuously and smartly monitor their health conditions.
Recent innovations are paving the way towards an even more interactive connectivity level with the ongoing developments towards the 5G cellular technology and the Tactile Internet, especially with the 5G massive machine type communications (mMTC) and ultra-reliable low latency communications (URLLC) use case scenarios. Combining extremely low latency with high availability and reliability, the Tactile Internet, based on 5G URLLC, will facilitate real-time, synchronous, haptic feedback with remote-control solutions. This will give rise to a great variety of opportunities for emerging technologies and services especially in the field of tele-health. With the prospective evolution towards next generation 5G networks, such services will benefit from faster connections, greater mobility, increased inter-device compatibility, and harmonization between different communications standards.
Remarkably, in sHealth systems for neurological applications, continuous sensory data acquisition and rapid diagnostic feedback and medical intervention have the potential to positively impact the lives of a large portion of the global population who suffer from chronic brain illnesses and disorders. Yet, these impose strict constraints on data storage, capabilities of networking technologies, processing speed, algorithmic complexity due to the high number of active sensors, the massive volume of collected biomedical data, and the need for premium service quality in terms of reliability and delay. For example, a wireless electroencephalography (EEG) headset that measures the electrical activity of the brain with 29 channels, sampled at 512 Hz with two bytes per channel, generates approximately 107 MB of data per hour for one single patient, which requires to be processed, transmitted, stored and analyzed. Despite the promising features of 5G in terms of large spectrum bandwidth and fast transmission, the challenge remains in how to leverage such features efficiently in terms of cost and energy consumption. This motivates the use of the edge computing paradigm, which promotes smart ways to process medical data close to where they are collected (i.e., as close as possible to the edge of the network), in order to optimize the amount and quality of data communicated over 5G, hence, minimizing the delay, energy consumption, and cost associated with data delivery.
A very important use case of using EEG measurements to monitor patients is epilepsy. In fact, characterized by recurrent seizures, epilepsy is a common brain disorder affecting around 1% of the global population. Potentially harmful consequences of epileptic seizures and the lack of a definite cure for a large number of patients encouraged alternative solutions that allow subjects to actively monitor their brain health through intelligent mobile sensing devices with seizure detection or prediction capabilities. EEG monitoring is the primary tool for the assessment of epileptic seizures and it involves analyzing the change in EEG activity before, during, and after seizure onsets. The purpose of seizure prediction is to alert the patient normally few minutes beforehand of a seizure occurrence in order to take precautionary measures. However, the problem of prediction based on EEG signals is highly challenging, especially in reducing the rate of false alarms. On the other hand, seizure detection tries to identify the onset of the seizure and possibly its offset. In this case, ultra-low delay is essential as a seizure can typically have a duration of few seconds and, thus, every fraction of a second matters. This constitutes an important and impactful, yet challenging, application scenario for 5G URLLC.
In this project, we present challenges and develop solutions for designing effective sHealth systems for neurological applications using a holistic framework that captures end-user sensing, mobile device processing, radio access network connectivity, mobile edge computing, remote cloud computing and storage, combined with advanced signal analysis and machine learning intelligence. We further illustrate the presented framework via a detailed experimental use case on mobile epileptic seizure detection and predication as an example of a globally common neurological disease. In addition, we investigate efficient resource allocation over 5G/5G+ networks in order to transmit the measured health information in real-time, and to alert the user in case of an imminent seizure. We showcase our achievements via a testbed based on real measurement data, and implemented in collaboration with Hamad Medical Center (HMC) and Huawei in Qatar.