Launching AI Solutions Beyond The Cloud
Today, we are taking a closer look at the cloud component. Some AI algorithms are developed in a cloud infrastructure, and strictly made available as a cloud solution. This service delivery method may not be acceptable for many healthcare organizations that cannot adopt it due to procedural or security concerns. As a vendor-neutral “broker” connecting caregivers and algorithm vendors, Dicom Systems is in a unique position to bridge the gap between cloud-based AI solutions and on-prem customers. We launched the AI Conductor to enable on-prem customers to implement AI solutions without disrupting existing workflows.
Reasons to Adopt The Cloud in Radiology
Cloud computing is facilitating greater integration and collaboration between hospitals, medical organizations, and healthcare providers. In a few short years, the healthcare cloud has expanded from functioning as simple data storage to enabling cutting-edge research and AI-based innovations.
The benefits of storing patient data and imaging studies in the cloud include:
- Enhanced security: As new levels of improved security are developed, cloud-based patient data storage is subject to continuous improvements with little or no disruption to the access of information.
- Scalability: Cloud storage can rapidly scale up or down, so as the data footprint changes, cloud storage adjusts accordingly.
- Reduced Costs: Storing patient data in the cloud means fewer resources are needed on-site, as these responsibilities are transferred to the cloud vendor.
- Availability and Reliability: In the cloud, data is safe and secure, with recurring power and backup power sources built into the system.
- Access: cloud storage allows access from any device, any time. The COVID-19 pandemic has accelerated the adoption of telemedicine and teleradiology, and the healthcare cloud has proven tremendously valuable in this endeavor.
Why Hospitals Choose To Keep Data “On-Prem”
Despite the many benefits of cloud-based medical data sharing, its adoption in radiology is far from ubiquitous. Due to a number of technical, security, and organizational challenges, many hospitals are choosing on-prem storage, at least for the time being.
A UK study on the Factors Limiting the Adoption of Cloud Computing in Teleradiology cites the following reasons hospitals are choosing to keep patient data on-prem:
- concerns about the reliability of their internet connection with the adoption of the cloud
- disaster recovery
- integration and interoperability
- cost, reimbursement, and insurance
- organizational culture, such as skepticism on the part of caregivers, some of whom still view AI as a risk to their patients, or a threat to their profession
- security concerns, trust, standards, and data privacy legislation
As an example, the study discussed a hospital that is taking extreme care to avoid a breach in patient data security. When it comes to sending data “off-site,” the line between the hospital and the cloud provider’s data center must be encrypted. This is an extra step that requires effort and cost on the part of the hospital, thus interfering with the adoption of the cloud.
Bringing AI Solutions to On-Prem Customers
At Dicom Systems, interoperability has always been a core guiding principle. To that end, our routing platform can be leveraged as a versatile AI workflow conductor, enabling the deployment of any AI on premises or in the cloud. The AI Conductor leverages industry standards to communicate with systems upstream and downstream along the imaging process continuum. By connecting via standard Docker, AI Conductor is able to integrate ‘n’ numbers of AI algorithms that can interchangeably reside on premises within the caregiver’s IT enterprise, or in a cloud infrastructure.
Looking Ahead: How To Vet and Select AI Solutions
Now that we have discussed the largest obstacles standing in the way of AI adoption, we can look forward to a more efficient AI ecosystem within the world of enterprise imaging.
One of the ways to facilitate faster mainstream adoption of AI is to identify early adopters and connect them with AI vendors. The ideal type of AI vendor for this phase will be ready to adapt their technology and make it easily adoptable in imaging workflows, whether on-prem or in the cloud.
We suggest the following key due diligence steps for vetting AI suppliers and customers:
- Is the developer of the AI the right developer? Does the developer have the competence, experience, and knowledge of imaging necessary to effectively build the intended algorithm?
- Does the AI vendor make effective use of industry standards?
- Has the AI vendor articulated a strategy for how to effectively deploy their algorithm into a wide variety of clinical enterprise IT ecosystems?
- Has the pricing and business model been validated on both sides of the equation? Is the pricing viable for long-term and sustained commercialization? Like any other technology, it’s important for customers to know that a vendor has the financial backing and viability to be a long-term partner. Conversely, has the healthcare customer validated that their budgets have room for the solutions?
These questions will assist in creating an environment of trust and reliability that each side must have in order to collaborate and gain meaningful results from AI.