According to a 2016 Johns Hopkins study, medical mistakes kill 250,000 Americans a year—behind only heart disease and cancer. Even if that number turns out to be too high, a recent article in The Wall Street Journal advises using a chatbot to help you “spot errors, understand lab reports and stick to care plans.”
In one experiment, a diagnostic chatbot solved medical cases reported in the New England Journal of Medicine with an 85 percent accuracy, roughly four times as well as primary-care doctors
using the same data. Because the chatbots “ingest” research from every
specialty, they can connect dots that doctors may miss.
Here's how the Journal article tells you to use a
chatbot:
Keep a health diary: Open your AI chat, such as
Claude, ChatGPT, Gemini, or other top model and insert your health information,
such as diagnoses, surgeries, lab results, current medicines and doses, and
even “mystery” symptoms. Ask the bot: “What patterns jump out? What’s overdue,
and which gaps should I address first?” If the answer is confusing, say,
“Explain that in simpler terms and give an example.”
Enlist AI’s analysis: Tell the chatbot about your
health concerns, such as a pain you’re experiencing. For example: “Find the
triggers to my pain flare-ups, rank likely culprits, and offer standard and
unconventional fixes.” AI might find a hidden side-effect from medication or a
disorder that doctors missed.
Clarify communications: If your doctor has recommended
a course of treatment or prescription that isn’t clear, you can ask the chatbot
for clarification. For example, if you don’t understand why an anti-anxiety
drug was prescribed for your stomach pain, the chatbot will explain: “The drug
calms overactive nerves.”
Ask your chatbot for second opinions: Seek
counter-evidence to any significant AI suggestion by asking: “Show
peer-reviewed studies—especially clinical trials—that argue against this
recommendation.” Assuming the studies will be dense, ask the chatbot to
summarize key points.
I've never done this, but it sounds sensible to me.
For an introduction to this blog, see I Just Say No; for a list of blog topics, click the Topics tab.
