Medicine has been one of the most renowned success stories of modern science and technology. However, the MIT Technology review observes that until 2020 the pace of digital transformation in this sector has been frustratingly low owing to its risk-averse nature and spiraling costs. The mainstreaming of digital tools for enabling the treatment outcomes was emerging, but slowly.
But the disruptions in the wake of the COVID-19 upended the scenario, stretching the global healthcare workforce to its limits. At the height of the pandemic, Mental Health America reported that stress and burnout affected 93% of the healthcare workers. It compelled the decision-makers to reconsider operational practices and find ways to build, manage and scale smarter hospital applications that intuitively assist the healthcare providers. Nevertheless, for such applications to deliver, the vast operational data streams need to be processed near their point of ingestion to reduce lags and enable real-time decisions.
Solving Real-World Problems
For instance, consider this use case. For patients undergoing treatment or residents of old age homes, falling from bed can be a significant issue, severely delaying recovery. In fact, research by Cleveland Clinic reports that 30% of such falls may result in serious injuries. However, continuous watch out across hospital wards can be extremely tasking for the medical staff. Here, can round-the-clock manual surveillance be replaced by bringing machine intelligence closer to the hospital floor? A smart application running object detection algorithms for face landmarks and body pose detection can predict a fall and trigger an alarm for the responders.
Such runtime feedback loops needed for remotely monitoring the patient’s body posture and vital signs and arriving at instant decisions based on situations are only possible by running AI algorithms on Edge.
What is Edge AI?
Currently, AI-powered solutions are implemented using powerful data centers on-premise or in the cloud. However, healthcare’s inherent challenges and peculiarities make this architecture difficult to be used across healthcare use cases. Running AI algorithms in the cloud comes with limitations like:
- Unreliable Internet Connectivity: While developed nations are way ahead in internet penetration and robust connectivity, it can be a challenge across the Global South. Further internet connectivity can be limited in rural areas and field hospitals.
- High Operational Costs: Sending data to the cloud and back to the device involves costs. Also transferring medical imaging in no-loss formats may push up operational expenses.
- High Latency: Putting massive data traffic through the internet can cause delays which is unacceptable in life and death situations. For instance, in the above illustration, the images of a patient’s position on the bed must be processed in real-time by the AI engine for optimized response.
In respite, Edge computing AI or TinyML brings the power into the device installed in the field. Instead of the cloud, the concept focuses on implementing neural networks at the endpoints or the network’s Edge. The AI-enabled edge device can thus process the data loads locally without relying on the cloud processing backend.
The Edge AI concept is based on the fact that the training and deployment of Machine Learning (ML) models can be done separately. Therefore it is possible to embed pre-trained ML models into medical devices with limited memory and computing, converting them into smart systems. However, challenges persist in efficiently handling AI workloads on the Edge owing to limited computational bandwidth, and the predominantly vendor/platform-specific nature of the available solutions.
Nevertheless, in the current digital economy characterized by the proliferation of the Industry 4.0 constructs and 5G, there is much optimism about Edge AI, with robust estimates from every corner. Research and Markets predict the global Edge AI software market will demonstrate a handsome CAGR of 19%, reaching nearly $2 billion by 2026. On the other hand, Edge AI Hardware Market Outlook — 2030 by Allied Market Research forecasts the market size for processors, memory, and sensors to reach $38 billion by the end of this decade.
Data-driven Healthcare
But what has led to the increased mainstreaming of the Edge-operated AI in healthcare in recent years? Apparently, an explosion of data in the sector due to the adoption of IoT and an increasing demand to harness it sustainably to deliver more personalized and intuitive clinical journeys. Dell Technologies reports that healthcare and life sciences presently account for 30% of all data stored globally, and about 3 million data points are generated on average across various clinical trials.
The data volumes are expected only to go up in the coming days due to the increased usage of connected devices and IoT sensors in healthcare. For instance, right now, at least 10–15 connected devices are in use per hospital bed in the US. Based on this trend, experts predict that 75% of the healthcare data will be generated at the Edge of the networks by 2025.
Benefits of Edge AI
Here Edge AI provides the tool to process the data near the source and bringing transformative benefits for healthcare like:
- Improved Security: Instead of data centers maintaining data within the Edge devices, confidential information on patient health remains secured from intrusions and less exposed to mass data breaches.
- Faster Triaging: Accurate diagnosis of health issues is key to delivering clinical outcomes and proper patient care. Here, operating alongside human healthcare professionals, AI solutions on the Edge can rapidly perform multiple tests at scale, giving better insights into the patients’ health. For instance, Google is leveraging AI to help doctors screen patients for diabetes-induced retinopathy and prevent early blindness.
- Lean Healthcare IT: Processing healthcare data using Edge AI allows healthcare institutions to adopt a leaner IT infrastructure. While operational and tactical aspects are pushed to the Edge, the cloud and data center bandwidths are focused on more strategic roles. Also, it ensures that vital healthcare processes are still available even in an outage.
- Process Automation: AI-enabled Edge devices can take on the repetitive tasks of the clinical environment and help healthcare workers to focus on more strategic tasks, saving time and money. For instance, in the US, on average, nurses spend up to 25% of their work hours on administrative works like patient onboarding and documentation. Instead, robotic process automation and Edge AI at the front desk can use tech like Natural Language Processing for initial patient interviews and capture and make the relevant information readily available for the healthcare professionals to review.
Edge AI Use Cases in Healthcare
While the benefits of inducting Edge AI in healthcare are apparent, what are some of the use cases where the technology is currently operating or may be adopted in the days ahead? Experts believe that the agility of Edge AI makes it highly contextual along the entire healthcare journey. Interventions include:
- First Response: Ambulances that ferry patients and accident victims to hospitals are no longer just transportations but slowly evolving into mobile Edge platforms that can deliver the necessary care within the golden hour. For instance, in Spain, EMS members use tablet PCs to capture patients’ vital signs and send them over a 5G network for analysis by the emergency personnel back at the hospital. In the days ahead, such information can be processed on the go using AI, directing the EMS professionals on the necessary steps to prevent the loss of life.
- In The Hospital: Within clinical environments, Edge computing and AI are pacing up the quality of diagnosis and automating medicine delivery. Instead of repeatedly transporting patients into various facilities for checks, Edge AI brings such services right to the patient. For instance, healthcare establishments like UCLA Health, Massachusetts General Hospital, or King’s College Hospital in London have inducted AI-powered MRI scanners that can operate at the patient’s bedside, identifying anomalies and helping radiologists analyze the situation in real-time. Further, instead of depending on nurses to administer the insulin shots to diabetics, an smart insulin pump can ingest data from artificial pancreas sensors under the patient’s skin to determine the blood sugar levels, automating delivery.
- At Home: One of Edge AI’s most prominent use cases in healthcare is telemedicine, delivering treatment directly to patients where they live. The American Medical Association and Wellness Council of America believe that upto 75% of the clinical workloads can be handled safely through telemedicine. For instance, using smart Edge sensors to monitor patient conditions at home can trigger alerts for the caregivers if the situation deteriorates. Also, Edge AI can help bring optimum healthcare to remote areas where quality medical expertise may not be available. The embedded intelligence can help process data locally and guide low-skilled medical professionals to make informed decisions.
Final Thoughts
While the benefits of Edge Ai in healthcare are multifaceted and compelling, much depends on skilled execution. In fact, in high-risk environments like healthcare, the imperative to get first-time-right outcomes can hardly be overemphasized! Therefore alongside investment in technology, it becomes a strategic necessity to find an experienced Edge and AI implementation partner who can get the job done and deliver the desired objectives.
Like other businesses, if you too are looking for low code development platforms Mindfire Solutions can be your partner of choice. We have a team of highly skilled and certified software professionals, who have developed many custom solutions for our global clients over the years.
Content Source: Medium
No comments:
Post a Comment