Agentic AI is steadily being implemented to support the radiologist workforce, but there are valid concerns and wariness surrounding AI usage in healthcare. Common fears include security risks, privacy concerns, and regulatory uncertainty. These concerns are reasonable when it comes to this new and rapidly evolving technology.

We cannot overlook the benefits, though. When regulated and used responsibly, agentic AI can improve reading turnaround times and ensure patients get life-saving diagnoses and care faster. It’s already being used across healthcare in a variety of modalities. According to the American Medical Association (AMA), 81% of physicians reported using some type of healthcare AI tool, with 76% reporting that AI had “at least some advantage” in patient care.

Completely avoiding AI is nearly impossible in the industry today. Rather, physicians, providers, and executives must do their due diligence to understand the successes and pitfalls of AI tools and make informed decisions for what is best for their specific team, practice, and patients. We owe a critical eye and fair evaluation of AI tools before completely writing off – or putting our full trust in – these solutions.
To reach the next level of agentic AI deployment in healthcare, executives and providers need to have a realistic idea of what to look for and avoid in AI-powered solutions and what questions to ask to find the right solution for their team and patients.

Questions about patient data and security risks
Potentially the most important factor when considering AI tools is how they handle the security of patient data. In the recent AMA survey, a majority of responding physicians agreed that AI tools’ patient data privacy (86%) and safety validation (88%) were their top concerns about and barriers to potential AI adoption. As healthcare-specific data protection laws, like HIPAA, have grown, AI solutions must be held to the same standards.

The healthcare industry is also a major target for cyberattacks and data breaches, regardless of AI implementation. Everything from patients’ personal information to imaging results can be targeted and released online. IBM reports a concerning trend of large breaches resulting in millions of records being stolen, and notes that healthcare sees the highest security breach costs, with an average of more than $10 million per incident.

When considering AI-powered solutions, healthcare executives should ask:
● What are the platform’s security measures? Have they been verified by third-party security experts such as HITRUST Alliance?

Artificial Intelligence Homepage