By Pascal Holt, Director of Marketing at Iceotope
Sustainability and our day-to-day health are becoming increasingly intertwined. Innovations in AI, digital infrastructure, and data security are determining the quality of our healthcare service, so how is it changing?
We all understand that entering the health system as a patient can be stressful. In the last 18 months, no matter what the condition, whether awaiting diagnosis, treatment, or a scheduled check-up the impact is acutely understood. Digital healthcare is at a pivotal point from the perspective of technological innovations facilitating improvements in disease treatment, improve individual well-being and personalised care.
Predictive care
Over the coming years, the evolution of healthcare will centre around the reengineering of clinical care and operations around digital health and pervasive, real-time use of data and advanced analytics to achieve these goals. Healthcare systems worldwide will be expected to deliver diagnostics and care that is both predictive and proactive, enabled by artificial intelligence (AI), machine learning (ML) and data-driven analytics, as connected care and bioinformatics commentators, such as the World Economic Forum annual meeting, forecast.
In the very near future, the application of advanced analytics, including AI and ML, will greatly improve clinical decision making and patient care outcomes. Analysing patient health records alongside vast datasets that cover populations, conditions, countries, environmental factors, virology data and more will be leveraged to help manage myriad health conditions.
Gartner believes, ‘Demands for care collaboration and coordination across the ecosystem are increasing the demand for real time data, insight and workflow optimization and orchestration.’ This is resulting in foundational technologies, such as the real-time health systems (Hype Cycle for RTHS Technologies). It predicts in coming years, ‘healthcare will be characterized by a reengineering of clinical care and operations around digital health and pervasive, real-time use of data to achieve goals.’
Understandably, healthcare providers and medical teams alike are excited about the potential for AI-powered diagnostics and precision medicine. Primarily, this is because of what this means for improvements in patient care – especially when so many countries expect the continuation of care for a larger population of senior citizens in years to come.
Data driven clinical informatics
Clinical informatics uses data and a range of tools to support health professionals. These include analysing data, preventing hospital patients having accidents on wards, running systems for storing and sharing X-rays, as well as ultrasound and magnetic resonance imaging (MRI) scans. Within a few years, AI will be used to access data sources and reveal patterns in disease, aiding treatment and patient care programmes. However, Health Education England forecasts that they will need to fill a skills gap of a staggering 672% to meet the expected demand by 2030.
AI-driven data analytics and resource-intensive task automation will enable healthcare providers, public and private, to increase productivity and efficiency of care delivery whilst enhancing resource use, reducing waiting times and tackling employee burn out. However, few medical or healthcare professionals have likely given much thought thus far to what these trends mean for the underlying critical IT infrastructure required to support this revolution.
Data demands will rely on higher density processing, especially high-density GPUs, which is evolving faster than the cooling technologies used within most data centers. The majority of PACS Administrators and IT departments have never seen data centres like these before, let alone have the capability to accommodate these requirements within their current IT infrastructure. This is taking place simultaneously with the phenomenon of ‘data gravity’, where the sheer volume of data being generated at the point of use is drawing the analytics, applications, and IT hardware to the data source itself, whether in operating theatres, on wards, or to the bedside for real-time processing. This is creating a whole new set of challenges to overcome.
‘A year in the life of the NHS AI Lab’, 2020, illustrated that diagnostics had the most prevalent use of AI within the NHS. This marks the beginning of using deep learning (DL), ML and categorisation technology on enormous sets of medical images to create workflows and algorithms. These allow for faster and more accurate outputs at the point-of-care. This means that an increasing amount of processing also needs to be done at the edge, closer to the point-of-care, giving rise to additional challenges such as available power, space, acoustics, as well as physical and data security.
Digital healthcare at the server level and sustainability targets
To facilitate change, strategic partnerships are required between healthcare providers, technology companies, data centres and associated organisations to drive towards digital transformation. Many in healthcare already see the positive results of investment in AI as a powerful enabler of operational efficiency, which leads to better diagnosis, treatment, and outcomes.
Much has been made of the data centre industry’s power inefficiency, add to that, the fact that the healthcare sector has its own sustainability challenges, producing the equivalent of 4.4% of global net emissions and you get greater IT infrastructure challenges. Healthcare leaders are set to prioritise sustainable initiatives, with projected cost savings as an additional driver, which many believe go hand-in-hand with technology advancements.
As the data centre sector continues to support most digital transformation, it is set to inherit a substantial amount of the sustainability challenges of other sectors. IT transformation, mobile devices, and Internet of things (IoT) are creating enormous volumes of data globally. IDC predicts that in 2025, 175 zettabytes (175 trillion gigabytes) of new data will be created around the world, while Gartner is forecasting that more than 50% of this data will be generated and processed outside of the data centre.
Transformational technologies, such as Digital Health Platforms (DHP), will enable healthcare providers to quickly respond to external uncertainty as well as planned change. They can do this using cloud-first healthcare applications and tools that encompass EHR, data connectivity and powerful analytics. In so doing, they can address strategic issues for providers, where monolithic EHR-centric application architecture fails to meet changing patient and clinical workforce demands. It is believed that DHP will reduce EHR total cost of ownership, release data for deeper insight and deliver higher clinical and cost outcomes.
In the coming years, the exponential upsurge in data processing necessary to extract patient insights from large datasets will continually drive the requirement for higher power compute densities and this is changing server cooling strategies. Data centres exist to process data as efficiently as possible to enhance the benefit for customers and owners.
CPU power consumption is on the rise, with Thermal Design Power (TDP) mapped to reach 400+ watts – resulting in hotter chips and higher rack densities. Increasing use of high-power GPUs alongside the CPU to accelerate computational workloads is also resulting in much higher power consumption and is driving the need for a fundamental review of thermal management in the data centre and at the edge.
Currently, the predominate way to remove heat from IT server equipment is by inefficient cold air drawn through the chassis, which now requires numerous internal fans to satisfy the higher density processors within the servers. Even the most efficient air-cooling systems cannot cope with the requirement of CPUs with mapped TDP of 400+ watts. Simply blowing more air over the problem is not practical, efficient, or sustainable. Efficiency of compute also requires the collaboration between servers, data centres, interconnectivity, and the customer to understand how best to move, process and store data. HPC and supercomputer level computations require specific layouts that increase the need for direct to component cooling.
Data centre PUE (the measure of energy efficiency) across the industry has plateaued over the past 5 years, and air cooled ITE is a significant brake on moving that needle towards PUE 1.2 or 1.3, which the leading colocation data centres are achieving. Liquid cooling onto the server motherboard provides the technology to affect an energy efficient strategy for the compute scalability that new healthcare applications require. Only through using the range of technology that removes computational barriers can we enable digital transformation both inside and outside the data centre to empower high-quality personalised healthcare delivery in the near future.
