Our AI Midwife

Using advanced AI for complex personal health questions can now provide critical insights missed by human doctors.

A couple struggling with unexplained infertility and multiple failed IVF embryo transfers utilized OpenAI's GPT-4 as an AI "doctor" to analyze their medical data. By prompting the AI to act as a character from a medical drama, they received detailed advice and diagnostic suggestions that were validated by their human fertility doctors. This AI, dubbed "Dr. Reid," eventually recommended an MRI of the wife's uterus, a procedure that had not been strongly suggested by their human physicians. The MRI revealed a large, previously undetected fibroid, which was subsequently surgically removed. Following the removal, the wife conceived naturally and gave birth to a healthy baby. The author reflects on the profound personal benefit derived from AI, contrasting it with the often negative public discourse surrounding the technology.

The author's experience highlights a significant gap in patient-doctor interaction, where complex or persistent issues may not receive the focused, iterative attention needed due to time constraints and system limitations. By leveraging LLMs with specific prompting techniques, patients can effectively augment their understanding and pursue more targeted diagnostic paths. This suggests a future where AI acts as a powerful, accessible patient advocate, capable of synthesizing vast amounts of data and asking critical questions that might otherwise be overlooked.

This case demonstrates AI's potential to democratize access to high-level diagnostic support, particularly for conditions that fall into "unexplained" categories. It also implies that the current regulatory and practice frameworks for healthcare may need to adapt to integrate AI-driven insights more formally. As LLMs become more sophisticated and accessible, individuals facing complex health challenges could increasingly rely on them to navigate the healthcare system and uncover previously hidden issues, potentially leading to better outcomes and reduced patient frustration.