Data-Focused Strategy May Hinder Generative AI Deployment in Healthcare

January 31, 2024 - Insights from Deloitte’s 2024 Health Care Generative AI Outlook Survey indicate that healthcare executives’ narrow focus on data could create blind spots that could prevent organizations from successfully leveraging generative artificial intelligence (AI).

The report surveyed 60 executives across the healthcare industry to learn more about their approaches to generative AI implementation in 2024.

Data considerations were top priorities for most, with 82 percent of executives reporting that they are focused on factors like data availability, reliability, and quality. Over 70 percent of respondents indicated that regulatory compliance and issues related to data privacy and security were also a major focus.

However, only 60 percent were focused on data governance, while 45 percent prioritized mitigating data biases.

The survey asserts that these responses describe a narrow, data-focused approach to generative AI in healthcare that may leave organizations vulnerable to major blind spots. These blind spots, defined as factors that less than 60 percent of leaders focused on, could hinder efforts to integrate generative AI in healthcare.

The first blind spot highlighted by the report is data governance. While it is a priority for some executives, the survey points out that the lack of focus in this area could damage both employee and consumer trust. Conversely, implementing a data governance framework has the potential to the strengthen trust by protecting patient privacy, reducing bias to advance health equity, and boost the effective use of high-quality data.

The second blind spot concerns the healthcare workforce. The report emphasizes that gaining employee buy-in is key to a successful generative AI strategy in the wake of the healthcare workforce crisis. Addressing concerns around the technology by promoting workforce literacy and focusing on generative AI as an assistive tool could help ease employees’ fears, but healthcare executives are less focused on these considerations.

Roughly 63 percent of healthcare leadership reported that using generative AI to reskill and upskill their workforce, rather than to replace them, was a major focus. Less than two-thirds indicated that they were prioritizing addressing employees’ concerns around the technology or providing change management as job roles and workforce composition shifts.

The next blind spot centers on healthcare consumers. The survey demonstrates that 50 percent or less of respondents are focused on building trust, ensuring equitable access to generative AI-based solutions, or educating patients on AI and its risks.

The survey underscores that the technology has the potential to significantly advance areas like clinical decision support and diagnostics, but demonstrating its value to healthcare consumers is the only way to achieve these aims. The report recommends that healthcare organizations prioritize consumer engagement, seeking insights into pain points and what AI solutions users are willing to utilize.

The final blind spot the survey highlights the challenge of scaling generative AI across the enterprise. Standalone application programming interfaces (APIs) can be difficult to scale, and a single model is unlikely to meet all of a healthcare organization’s needs. The report notes that there are numerous operational and technical hurdles – like deploying strong data pipelines and managing vector storage – stakeholders must contend with.

To that end, the survey suggests that implementing machine learning operations (MLOps) capabilities can help improve generative AI scalability upfront.

“By addressing consumer and workforce considerations alongside the data considerations, health care organizations can pave the way for a future in which generative AI not only augments health care delivery but does so equitably, without bias, in a trustworthy and ethical way, along with a personal touch,” the report stated.

These findings provide a clearer picture into the complicated landscape of generative AI’s role in healthcare.

In an interview this month with HealthITAnalytics, leaders from Forrester, PwC, and Duke AI Health provided predictions for what the AI and analytics space is likely to look like this year. They indicated that the healthcare industry’s interest in generative AI will continue, as its potential for use cases like generating discharge summaries, improving the prior authorization process, searching electronic health records (EHRs) for patient information is attractive.

However, they emphasized that successful deployment of the technology will rely on healthcare organizations’ ability to mitigate risks related to AI bias and trust.