Using ChatGPT before a medical appointment? How to get the best support for your health questions
Chatbot advice is readily available, is personalised and can be prompted for greater clarity. But it can also lead to serious patient harm.
Erin, a 40-something woman, is feeling exhausted, irritable and experiencing cognitive decline. She suspects her symptoms may be due to perimenopause. Seeking advice, she searches ChatGPT for information about the condition and treatment options. She is not alone; between 30% and 55% of adults have used AI chatbots for health inquiries, with usage increasing.
The convenience of chatbots, especially outside traditional healthcare hours, makes them appealing to many. However, the use of AI chatbots for health raises concerns regarding patient safety and care quality.
Many people turn to chatbots to satisfy their curiosity or decide whether to see a GP. Around half find chatbots more convenient than regular services, often seeking assistance late at night or early in the morning when traditional healthcare services are unavailable. Over half of those who used chatbots believed they positively impacted their physical and mental health. A study analyzing over 500,000 health conversations using Microsoft's co-pilot found one in five related to personal symptoms and treatments.
While chatbots may provide quick, personalized information, there are risks associated with their use, particularly when individuals need urgent medical attention. Approximately 20% of respondents in a UK survey decided against seeking medical advice after consulting a chatbot. Additionally, a study examining AI responses to 222 patient questions relevant to primary care found that between 5% and 13% of the responses were unsafe, potentially leading to serious patient harm.
Furthermore, when chatbots were used by participants, accuracy dropped significantly, with only 34.5% accuracy.
Chatbots' effectiveness is also limited; they may not be as reliable as web searches for understanding underlying conditions, and only 43% accuracy when deciding whether to seek professional care. A key issue is that people often do not provide sufficient details in their prompts to obtain accurate results, and this process requires striking a balance between specificity and privacy.
Moreover, consumers often fail to verify the information provided by AI chatbots, with only 1% clicking on the links provided by AI Google summaries.
While patients and clinicians could collaborate to discuss chatbot use and support shared decision-making, many do not disclose their chatbot usage. Reasons for this reluctance include fear of a negative reaction or concerns that clinicians might feel undermined. However, recent US data suggests that around 70% of clinicians view patient AI use for general health information or decision-making positively or non-disruptive to practice.
To improve outcomes, disclosing chatbot use enables clinicians to correct inaccuracies and engage in shared decision-making. However, clinicians report being poorly trained and lacking time and reimbursement to support patients' AI use effectively. While some resources are available, patients require more guidance in using chatbots, and clinicians need better guidance on how and where to direct patients for support.
Written by urgent.news from The Conversation AU's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
