Dr. Erkan Hassan

Monthly Writings

Evaluations and reviews of the latest in the field.

Scaling AI Tools Across an Enterprise Health System

After reading this article, you will be able to describe the risks of AI scaling from a pilot site to an enterprise health system deployment and how to avoid them for success.

SUMMARY:

  • Scaling is an important step in determining the success or failure of an AI tool.

  • Scaling from a single pilot to an enterprise program rarely translates seamlessly

  • Strategic planning is needed to achieve a sustained, safe, effective, and consistent outcome.

CONSIDERATIONS

  • Scaling ≠ Copy and Paste

  • Scaling is the deployment of new tools for enterprise planning and design, coupled with local adoption.

  • Implementing an enterprise AI solution is NOT about the technology but rather an overarching strategy of integration and trust.

REQUIREMENTS FOR SCALING AI TOOLS

  • Treating the scaling of AI tools as an organizational strategy

  • Proven evidence of safety and clinical impact

  • Developed a framework to improve consistently:

    • Operational efficiency

    • Patient outcomes

    • Evidence-driven innovation

  • Improve clinician workflow

  • Clinical Champion

  • Continuous monitoring

SCALING PATHWAYS

  • There are 2 forms of AI tool scaling:

    • Expand the usage of an already developed tool

    • Expand the number of AI tools to current sites

  • Each pathway has a different goal and set of risks

SUCCESSFUL SCALING STRATEGY 

  • Scaling is a people challenge before it is a technology challenge.

    • Constant training on local issues

    • Optimal use of clinical champions as peer educators and problem solvers.

  • Determine an Enterprise scaling strategy

    • Systemwide AI tool needs

    • Learn the successes from pilots

    • Correct what did not work

    • Validate interoperability early

  • Introduce AI tools deliberately

    • Determine interoperability early

    • Sequence deployment intelligently

  • Prioritize sites based on strategic value, operational readiness, and patient needs

  • Clinician cognitive load increases with multiple tools

    • Establish a detailed training plan with integration and escalation points.

  • Monitor Performance

    • Review and re-establish parameters at each site

    • Compare pilot results at each site

    • Develop an easy-to-use monitoring dashboard

    • Detect inconsistencies early

    • Track over time

    • Share findings

    • Update training and processes as needed

  • Vendor relationships evolve as you scale

    • As the number of vendors increases, oversight needs must scale too.

CONCLUSIONS:

Successful scaling of an AI tool is fraught with risks.  An enterprise strategic plan is needed for the innovative system to be successful.

  • Success will not translate to outcomes automatically

  • Implementation is NOT the goal; rather, sustained, effective, and consistent performance is the desired outcome

  • When done properly, you accelerate your organization into a leadership position.

  • Adequate AI tool scaling is an important step in the success or failure of a project.

  • Scaling is not copy-and-paste from the pilot site to each clinical site.

  • Scaling of AI tools requires an organizational strategy and is fraught with risks.

  • A successful scaling strategy is presented.

    Let’s have a brief chat to discuss your unique situation

Erkan Hassan