Socio-Technical Factors Driving Success in AI and Data Healthcare Initiatives
The Crucial Conversations series provides dynamic opportunity for participants to hear diverse perspectives and insights from an expert panel with varying vantage points. Attendees also participate in collaborative discussion on what they understood, share their own unique perspective and talk about ways we could collectively address challenges and opportunities in care moving forward. Through curated small group conversations, participants in these sessions unpack the complexity of our industry and think about realistic steps that can be taken to make improvements, even incremental ones.
Socio-Technical Factors Impacting AI and Data Initiatives in Healthcare
In our recent AI is Nothing Without Data Crucial Conversation our expert panel, comprised of Sherri Zink, SVP & Chief Data Engineer, BCBST, Dr. Lynn Simon, President, Healthcare Innovation & Chief Medical Officer, CHS and Tammy Hawes, Founder & CEO, Virsys12, moderated by Amy Deaton, COO of EvidenceCare, discussed the critical role of socio-technical factors—such as user training, organizational culture, and technology infrastructure—in the success of AI initiatives in healthcare.
How Culture Impacts Technology Adoption in Healthcare
Tammy Hawes, Founder & CEO, Virsys12, highlighted the need to foster a culture that is open, ready and willing to adopt innovation - like any technology or business strategy set into motion. Notably, this culture needs to be a trusting culture of accountability. In early stages of implementation there is a need to adjust and course correct data (as required), making it imperative to support a culture focused on not introducing any bias into the data. Additionally, strategic change management can make or break the success of an adoption. Many organizations that fail in their adoption of new technology, be it AI or another technological innovation, do so by not having a meaningful change management strategy or culture of accountability.
Government regulations, as we move towards a higher utilization and application of AI in healthcare, will either facilitate or hinder AI adoption. There is an opportunity for government regulations to enhance healthcare's ability to adopt AI efficiencies.
Key Takeaways:
- Cultural Readiness: The right organizational culture is crucial for adopting AI. This includes a trusting environment and a culture of accountability, especially important during the early stages of AI implementation to address and correct biases in data.
- Government Regulation: Regulatory frameworks can either facilitate or hinder AI adoption. For example, discrepancies in provider databases due to outdated information can complicate AI's role in healthcare.
- Strategic Initiatives: AI implementation should be treated like any other strategic initiative, with strong change management practices to ensure adoption and accountability.
- Healthcare has historically struggled with data sharing, which is essential for smart AI models that need comprehensive information from all sources involved in patient care. Efforts are underway to improve data sharing practices across the industry. For AI to be effective, it must incorporate data from multiple organizations, not just one. The motivation to share data is increasing as the benefits of comprehensive data integration become clearer to all stakeholders.
Key Takeaway:
- Shared Data for Smarter AI: AI models in healthcare need access to comprehensive data from multiple stakeholders to be truly effective.
Provider Response: People, Workflows and Infrastructure
On the provider side, Dr. Lynn Simon, President, Healthcare Innovation & Chief Medical Officer, CHS, explained when trying to deploy technology involves changes in workflow and how these changes can be incorporated into the existing workflow. Each interaction represents an important component in the chain of patient care, making adoption slower as it is critical to understand each use case, each deployment and so on. When these changes affect patient care or important decisions surrounding patient care it is imperative not to move too fast, but rather with diligence and pragmatism.
Key Takeaways:
- Technology vs. Change Management: Successful AI deployment is more about change management (80%) than the technology itself (20%). The main challenges lie in integrating AI into existing workflows.
- Human in the Loop: In clinical settings, AI should augment decision-making rather than replace it, ensuring a caregiver remains involved to provide oversight.
When it comes to challenges related to disparate systems and pulling large amounts of data into a single repository, there is a high level of governance required to normalize the data in a meaningful manner - the key comes down to infrastructure. To execute successfully organizations need the right infrastructure, technical capabilities and/or partners (if necessary). The data has to be normalized, definitions must be agreed upon, and there needs to be a solid governance process as it is being built. Once the infrastructure systems have been set into place, the next challenge becomes maintaining those systems over time.
Key Takeaways:
- Data Challenges: Integrating diverse data sources is critical. Organizations must normalize data from different EMRs and clinical sources to create a usable clinical data warehouse.
- Partnering for Solutions: Collaborating with partners to create a clinical data warehouse can help normalize data from multiple sources, which is essential for making data usable in patient care decisions.
- Governance and Maintenance: Strong governance and agreed-upon definitions are necessary to maintain the integrity of data systems over time
Alignment and Relational AI Data Structures
Sherri Zink, SVP & Chief Data Engineer, BCBST underscored the importance of technological change management initiatives to be aligned with corporate culture and goals. Regardless of the solution and its potential advancements, if your culture is not positioned to adopt and the model is not integrated into the workflow then it has high potential to fail.
Key Takeaways:
- Alignment with Business Goals: For AI initiatives to be embraced, they must align with the organization’s business goals.
- Workflow Integration: AI needs to be seamlessly integrated into existing workflows. This requires developing models that directly support daily operations.
Sherry highlighted the need to integrate clinical data. BCBST creates a supplemental database that converts clinical information into a claims-like format, ensuring compatibility with both upstream and downstream systems. This process emphasizes the need for portability, configurability, and ease of maintenance. As we look to the future, consolidating both structured and unstructured data will be crucial, particularly for leveraging generative AI to manage and classify documents across the organization effectively. This evolution towards normalizing unstructured data is significantly impacting our capabilities.
Key Takeaways:
- Structured vs. Unstructured Data: Current efforts focus on relational and structured databases, but the future will require managing unstructured data (e.g., documents, text) for generative AI applications.
- Master Data Management: Effective data sharing and master data management strategies are essential for AI models to function optimally, requiring collaboration across organizations to share patient data.
AI is Nothing Without Data Integration and Infrastructure
The evolving landscape of data sharing in healthcare will play a pivotal role in advancing AI capabilities. Overall, the panel highlighted that the success of AI initiatives in healthcare hinges on a combination of cultural readiness, robust change management, strategic alignment, and advanced data integration techniques.
In our next "AI is Nothing without Data" Crucial Conversations recap we will dig into data sharing practices and security with our expert panel: Sherri Zink, SVP & Chief Data Engineer, BCBST, Dr. Lynn Simon, President, Healthcare Innovation & Chief Medical Officer, CHS and Tammy Hawes, Founder & CEO, Virsys12, moderated by Amy Deaton, COO of EvidenceCare.
Join the Next Installment in the Nashville Health Care Council's Crucial Conversations Series
The conversations continue November 13, 2024 on the crucial topic of workforce. Members of the Nashville Health Care Council can register today. Not yet a member? Join our experienced and robust ecosystem of healthcare leaders.
The Nashville Health Care Council strengthens and elevates Nashville as the Healthcare City. With a $68 billion economic impact and 333,000 jobs locally, Nashville’s healthcare ecosystem is a world-class healthcare hub. Founded in 1995, the Council serves as the common ground for the city’s vibrant healthcare cluster. The Council offers engagement opportunities where the industry’s most influential executives come together to exchange ideas, share solutions, build businesses, and grow leaders.