Evaluate internal resources for adoption

Why does this matter?

Adopting and managing an AI product is a resource-intensive commitment. It creates a significant opportunity cost to not pursue alternative problems and solutions. Thus, it is important to have a methodology to identify necessary resources before making the commitment. 

• Standardize the process. It helps a variety of different functions have valid considerations for identifying and allocating the specific type and quantity of resources needed to adopt each AI product. 
• Make it easy to cut losses in future steps. It helps an organization mitigate resource waste on AI products that fail to realize their potential due to a lack of adequate support. 
• Know where the organization is at. It helps an organization understand its AI maturity, including considerations related to scalable solutions, integration and interoperability challenges, and a trained workforce. 

How to do this?

Step 1: Assess required resources

• Form a team to assess resource needs. This team should include a clinical champion who is a respected leader in the proposed context of use; an administrator who is responsible for the quality of care delivered within the proposed context of use; a data scientist with familiarity in the modeling approach used by the AI product; and an IT leader who is familiar with the operational technology systems that are used in the proposed context of use. If there is no internal data scientist to help assess the AI product, consider hiring an external consultant.
• With the representative team, assess resources needed to adopt, integrate, and maintain a specific AI product, including funding, staff expertise, and digital infrastructure.
• When assessing needs, consider each stage of the AI product lifecycle, including post-deployment monitoringContinual checking, supervising, critically observing or determining the status in order to identify change from the performance level required or expected. and alignment to applicable regulatory requirements. For example, understand whether the AI product is an FDAFood And Drug Administration Is A Government Agency That Regulates The Safety And Effectiveness Of Food, Drugs, Medical Devices, And Other Products.-regulated device. 

Step 2: Assess current capabilities

• Each healthcare organization has different environments, workforce structures, departments, and expertise available. Seek reliable information about existing capabilities and resources.
• Document self-assessment of current capabilities for adopting and using AI products. 

“…finding dollars for your core infrastructure and digital data environment. Whether you buy that from somebody else, whether you develop and maintain it yourself, it’s something that AI models; building, testing, deploying, running, you cannot do it without a really strong digital data environment…”

Operational Leader

Step 3: Identify gaps

• Assess the gap between existing capabilities and required resources based on the assessments conducted in the previous steps. This assessment helps your organization better understand gaps that need to be filled to successfully implement the AI product. 
• Be aware that addressing the gaps may necessitate new positions and appointments internally or the identification of external partners or solutions. The choice of resourcing will largely be influenced by the desired scale of AI adoption across the organization and the digital maturity of the healthcare delivery organization.

Step 4: Review for alignment to organizational needs and capabilities

• Determine whether the identified new capabilities needed to solve the problem (identified in the decision point on identifying and prioritizing a problem) are worth the cost of using an AI product. 

Step 5: Request ongoing support

• Document the assessment in a format acceptable by the institution to request the appropriate resources to support the AI product throughout its intended lifetime. Ensure that there are adequate resources to design and test a new optimal workflow, generate evidence of AI product safety and efficacy prior to clinical integration, execute the clinical rollout, and monitor and maintain the AI product.