Guides

Adapt to enable AI implementation

Why does this matter?

It may be necessary to tackle organizational change to enable successful AI product integration. Organizational changes may include social interactions or interventions that challenge existing structures, workflows, and technology capabilities. AI product integration may also necessitate the formation of new partnerships to coordinate with outpatient centers, insurance companies or other healthcare services, depending on the AI scope of purpose and workflow. Failure to adapt organizational structure will undermine the success of the AI product. 

How to do this?

Step 1: Survey existing conditions

  • Learn about the internal and external interests that benefit from the status quo and may resist AI product integration. 
  • Evaluate IT systems for necessary infrastructure for AI product integration.

“If we can simplify the output of the model down to a single number or a single color code or something and put in one small place as an epic user interface element, that’s what we should do. We don’t want to over complicate things.”

Technical Leader

Step 2: Work with senior leadership to drive organizational change

  • Meet with senior leaders to set expectations about organizational change that may be required for successful AI product integration. This may include:
    • Creating new structures or clinical positions
    • Changing existing roles
    • Migrating away from a legacy technology 
  • Ensure senior leadership buy-in for AI product integration to support organizational change.

Step 3: Map the organization chart and funds flow within the clinical integration context

  • Build an organization chart and funds flow diagram for the business units most affected clinical service lines. This step should be completed after building the patient journey map for the care pathway affected by the AI product (see guide design and test workflow for clinicians). The chart and diagram should include:
    • All clinicians affected by the AI product workflow 
    • Targeted end-users 
    • Performance metrics for affected clinicians 
    • Promotion criteria for affected clinicians
    • Reporting relationships for affected clinicians, business unit managers, and technology leaders
    • Revenue drivers for the clinical service line
    • Cost centers for the clinical service line

Step 4: Align incentives for end-users to facilitate successful integration

  • Identify potential challenges from below (adapted from this article) and take necessary steps to align frontline clinicians with objectives set for successful use of AI product.
AI Implementation ChallengesSolutions for Overcoming Implementation Challenges
The tool might deliver benefits to third parties rather than targeted clinician end usersIncrease end user benefit
  • Identify the clinician end users’ pain points
  • Develop a suite of interventions that address clinician end users’ patin points
  • Increase incentives for clinician end users to accomplish the outcomes the AI tool is designed to improve
The tool might require labor by clinician end users who are not the primary beneficiaries of the toolReduce clinician end user labor:
  • Engage third-party stakeholders in data construction, and start with currently available, smaller data sets
  • Engage third-party stakeholders in model evaluation, and simplify evaluation steps for clinician end users
  • Engage third-party stakeholders to take on some of the work required for tool use, and simplify the user interface
The tool might curtail the autonomy of clinician end usersProtect clinician end user autonomy:
  • Protect the tasks that clinician end users see as core to their work
  • Ensure clinician end users help evaluate the tool
  • Involve clinician end users throughout the AI product lifecycle

Step 5: Compensate for the labor required to integrate the AI product

  • Compensate effort and time required from end-users and affected clinicians for their role in AI product integration.
  • Identify a clinical champion from each relevant clinician stakeholder group who will have a formal role to serve as an advocate for and support their peers during AI product integration.

Step 6: Reward end-users and affected clinicians for improving patient outcomes

  • Create incentives to promote use of the AI product to address the problem. Incentives may be in multiple forms including:
    • Monetary 
    • Performance measures 
    • Promotion criteria

“Build that workflow [that never existed] before you can implement the technology… essentially you’re taking the technology and the workflow and operational process and making it all cohesive.”

Technical Leader

adopt health ai