Nowadays, the IT departments of many medical institutions are suffering from data storage and network speed. Doctors diagnose patients to help them get out of trouble. This treatment process requires good management.
The continuous spread of the coronavirus epidemic has intensified the rapid transformation of the infrastructure of medical institutions, which has adversely affected their business and data processing.
The widespread problem of data storage has emerged in many industries. For example, a data gravity index analysis report released by the data center operator Digital Realty pointed out that if a large amount of data cannot be better stored, it will affect the business of various industry organizations and bring resistance to change.
Remote work becomes the norm
Due to the outbreak of the epidemic, many companies and institutions have adopted the remote working model. Research shows that the trend of remote work will not be reversed. According to the forecast of Juniper Research, a research organization, a permanent impact of the epidemic on medical institutions is that the market demand for remote consultation and digital therapy will increase by 69% every year. The company expects that market revenue in this area will increase by 2025 and may reach 53.4 billion US dollars. Due to the massive growth of medical research data, drug production developers such as GlaxoSmithKline have built supercomputers in Europe with a speed of up to 400 petaflops in order to accelerate drug research to accelerate drug research.
Organizations such as the International Coronavirus Data Alliance (ICODA) stated at a seminar held in March 2021 that companies need to treat IT infrastructure issues (rather than rapidly growing data volumes) as the main challenge for their business development.
Edge computing technology cannot be a panacea for medical institutions to solve infrastructure problems, because the medical industry is a service category with a long history, and the challenges it faces vary with culture and infrastructure. For example, in the British healthcare system, management culture may be more difficult to reconcile than electricity, cooling or communication connections.
Although many companies and institutions have data center operations and management teams with clear responsibilities, IT staff in medical institutions may not be proficient in technologies such as power supply, cooling, and communication racks. If the hosting service provider can help them, then this is not a problem, but building the required infrastructure is a huge challenge.
Find the deployment room
A hosting service provider said that just deploying a micro data center in a hospital is a problem. The service provider is committed to deploying edge computing IT facilities in the hospital. The company said, “The most difficult part of our edge computing work is to find a room in the hospital building suitable for deploying IT equipment.”
The biggest challenge it faces is time constraints. The hospital IT project leader may decide that the server needs to be deployed immediately, but the IT provider may take three months to complete the deployment.
For example, in a project, the hospital appointed an installer to build a micro data center. The hosting service provider suggests the best option to the installer and waits for instructions while processing the report. During the initial inspection, the installer found a room that is very suitable for deploying a colocation data center. After a few months, the hospital switched to another room and had to be redeployed, but it would take 8 months to complete these tasks. When everything was ready, news came that the building was about to be demolished.
An installer who did not want to be named said: “To build a data center in a hospital, we need to think twice and find a room suitable for installing data center equipment, but if we can’t provide suitable space, we can only install it in the parking lot or the corner of the room. Or deploy.”
Monitoring and cooling
There is usually an installation mode for deployment in hospitals: build one or two data centers (one is the main business data center, and the other serves as a disaster recovery facility), where there are multiple independent power supply rooms, and each power supply room is not installed. Discontinuous power supply (UPS). The hospital uses 200 small UPS power supplies, which are scattered and difficult to manage. The hosting service provider can only remotely monitor through the built-in DCIM platform and aggregate all the information to the client.
As part of the edge computing facility, providing container data centers that use liquid cooling can solve multiple problems in hospitals. Since the server is immersed in an insulating cooling liquid, it can prevent the possibility of a fire caused by overheating of the circuit. The immersion cooling method can also prevent dust from being blown onto the circuit board, thereby avoiding short circuits.
In private healthcare organizations with faster decision-making speeds, the challenge of deploying IT equipment is how to fully adapt to the environment. Nursing homes are another example of how edge computing can solve network congestion and compliance issues.
The epidemic has put tremendous pressure on the staff of the nursing home, making it more difficult for them to spare time to take care of the sick. At the same time, they have to face new privacy regulations that limit the number of cameras that can be installed for surveillance.
The University of Amsterdam (UoA) proposed a way to solve this problem. Dr. Harro Stokman invented a way to use artificial intelligence to understand the pattern of events in each room.
Privacy regulations limit the time staff can view patients through cameras. However, there are no such restrictions on computers, and if they are considered smart enough, the health of patients can be judged based on these computers. This is the logic of the KNN artificial intelligence system developed by Dr. Stokman, which observes patients and determines whether human intervention or care (such as a fall) is needed.
The problem is that the KNN system creates too much data, and sending all of it to the cloud platform for storage and processing may cause network bottlenecks and more costs. In response, a company called Kepler Vision Technologies (KVT), a spin-off of the University of Amsterdam (UoA), used Nvidia’s Jetson Xavier NX module to build an edge computing device Edge Box, which can process all data locally and improve the collected data quality. Through localized analysis, the data sent to the cloud platform for processing can be reduced.
Still needs supporting network infrastructure
Creating edge computing hardware is one thing. But where does the supporting network infrastructure come from?
Dean Bubley, founder of the research firm Disruptive Analytics and a mobile telecom industry observer, warned that people’s expectations for the adoption of 5G technology are unrealistic, especially in support systems that require immediate response time.
Bubley said that in some cases, the ultra-reliable low latency (URLLC) associated with 5G could minimize the network round-trip time of new applications and devices that require immediate response. He said, “In this regard, mobile edge computing can meet their needs in the form of regional computing facilities or servers for each base station.”
But in many applications, this network latency must be lower. Endoscopes or microsurgery tools may need to respond to tactile feedback sent 100 times per second. Others have proposed to use drones for drug transport between hospitals, but these drones must respond to control signals or risks within two milliseconds. Whether 5G can provide the waiting time required for research photon sensors is also questionable, because photon sensors need to operate in picoseconds.
Can fiber optics help?
One of the ways the United States has responded to infrastructure challenges is to open access or competitive fiber optic networks, such as SiFi Networks’ FiberCity fiber optic products. The company promises that this will allow access to multiple service providers and different paths on the fiber optic network, thereby providing 99.9999% reliability.
In this model, every home user and enterprise will adopt a city-wide fiber optic network and effectively provide them with a dedicated network. Hospitals or research institutions can also transmit data at high speeds via optical fibers.
Ben Bawtree-Jobson, CEO of SiFi Networks, said that this would significantly reduce the problem of sending data through the cloud platform. Storing files in local facilities could also cause problems and needed to be shared off-site or internationally. Cloud storage might be more suitable for such seamless collaboration.
Bawtree-Jobson said: “The question is how much bandwidth is needed, and the use of fiber optic networks can solve this problem.”
At the same time, according to a survey report released by Juniper Research, many mobile operators are establishing partnerships around the world to build mobile edge computing infrastructure. By 2025, well-known manufacturers such as AT&T in the United States and LG Google in South Korea will invest US$8.3 billion to build network facilities for these edge computing systems.