The Rise of AI in Patient Health Management: A Double-Edged Sword
In an era where technology permeates every aspect of life, the healthcare sector is witnessing a significant transformation. Patients are increasingly turning to artificial intelligence (AI) tools to interpret their medical data, a trend that raises both opportunities and concerns. This shift is exemplified by the experience of Judith Miller, a 76-year-old resident of Milwaukee, who utilized an AI assistant to analyze her lab results after receiving an alert about them online.
The New Age of Patient Empowerment
Miller’s journey began with routine blood work in July. After receiving her results, she reached out to her doctor for clarification on some abnormalities, specifically elevated carbon dioxide levels and a low anion gap. While waiting for a response, she decided to input her test results into Claude, an AI model developed by Anthropic. “Claude helped give me a clear understanding of the abnormalities,” Miller noted, emphasizing that the AI’s assessment alleviated her anxiety during the waiting period.
This scenario is becoming increasingly common as patients gain unprecedented access to their medical records through online portals like MyChart. Federal regulations mandate that healthcare organizations promptly release electronic health information, including test results and physician notes. A 2023 study revealed that 96% of patients desire immediate access to their records, even before their healthcare providers have reviewed them.
The Role of AI in Health Information
As patients navigate their health data, many are turning to large language models (LLMs) such as OpenAI’s ChatGPT, Claude, and Google’s Gemini for assistance. These AI tools can provide insights and interpretations of medical information, but they come with inherent risks. Medical professionals and patient advocates caution that AI chatbots can generate inaccurate information, potentially leading to misunderstandings about one’s health.
A 2024 poll conducted by the Kaiser Family Foundation (KFF) found that 56% of adults who interact with AI are skeptical about the accuracy of the information provided. Adam Rodman, an internist at Beth Israel Deaconess Medical Center and chair of a generative AI steering group at Harvard Medical School, remarked, “LLMs are theoretically very powerful and they can give great advice, but they can also give truly terrible advice depending on how they’re prompted.”
Navigating the Risks of AI
The challenge lies in the fact that many patients lack the medical training necessary to discern whether the information provided by AI is accurate. Justin Honce, a neuroradiologist at UCHealth in Colorado, emphasized the need for caution when using AI tools. “Ultimately, it’s just the need for caution overall with LLMs,” he stated, acknowledging that while advancements are being made, concerns about accuracy and reliability persist.
Interestingly, the use of AI in healthcare is not entirely new. Patients have long sought information online, utilizing resources like WebMD and PubMed to research their conditions. However, the ability of AI chatbots to generate personalized recommendations and second opinions in real-time marks a significant evolution in patient engagement.
The Importance of Digital Health Literacy
Liz Salmi, communications and patient initiatives director at OpenNotes, has been investigating the effectiveness of AI in interpreting medical data. In a recent proof-of-concept study, Salmi and her colleagues assessed the accuracy of responses from ChatGPT, Claude, and Gemini to patient inquiries about clinical notes. The results were promising, but Salmi noted that the way patients frame their questions significantly impacts the accuracy of the AI’s responses.
Privacy concerns also loom large in this new landscape. Salmi cautioned that patients should avoid sharing personal information, such as names or Social Security numbers, when interacting with AI models. Data shared with these platforms often goes directly to the tech companies that developed them, raising questions about compliance with federal privacy laws. Sam Altman, CEO of OpenAI, recently warned against inputting personal information into AI systems.
The Need for a New Approach
As patients increasingly rely on AI for health information, a new form of digital health literacy is essential. This includes not only knowing how to ask questions effectively but also verifying AI-generated responses with other sources, consulting healthcare professionals, and safeguarding personal information online. Salmi and Dave deBronkart, a cancer survivor and patient advocate, emphasize the importance of these skills in navigating the complexities of AI in healthcare.
Healthcare providers are also exploring the use of AI to enhance patient communication. For instance, Stanford Health Care has developed an AI assistant to help physicians draft interpretations of clinical tests and lab results for their patients. A study conducted by Colorado researchers found that ChatGPT-generated summaries of radiology reports improved patient understanding, although some patients reported confusion due to the AI’s tendency to overemphasize or underemphasize findings.
A Personal Journey to Understanding
Returning to Judith Miller, after several follow-up messages through MyChart, her doctor ordered additional tests, which ultimately returned normal results. Miller expressed relief and noted that her experience with AI had made her more informed about her health. “It’s a very important tool in that regard,” she said. “It helps me organize my questions and do my research and level the playing field.”
Conclusion
The integration of AI into healthcare represents a paradigm shift in how patients engage with their medical information. While the potential benefits are significant, including enhanced understanding and empowerment, the risks associated with misinformation and privacy concerns cannot be overlooked. As patients like Judith Miller navigate this new landscape, the importance of digital health literacy and cautious engagement with AI tools becomes increasingly clear. The future of healthcare may well depend on finding the right balance between technology and human oversight, ensuring that patients can make informed decisions about their health.
