When I was studying genetic counseling (many years ago now), I learned early on how important it is to convey risk clearly to patients and how difficult it can be.
The way risk is presented can nudge people in one direction or another, depending on how they interpret it. The same risk, framed differently or given different context, can lead to very different decisions:
“That seems really high, so I’m not sure I want the procedure,” versus “That’s not a huge risk, so I’m seriously considering it.”
Same risk, different interpretation, different decision.
Add AI to the mix, and there’s an even greater danger of oversimplifying, inserting bias, or influencing patients’ decisions inadvertently.
AI doesn’t handle uncertainty particularly well. It likes to present information in ways that sound clear, confident, and definite. So when it’s left to its own devices, it may explain risk in ways that aren’t fully accurate or helpful to the reader.
That said, you don’t have to avoid AI entirely. Here are a few thoughtful ways AI can help you present risk:
- Make numbers interpretable: Turn abstract statistics into formats people can grasp more easily.
Example: “A 2% risk” → “2 out of every 100 people experience this.”
Prompt: “Rewrite this risk so a general audience can understand it. Include both percentage and ‘X out of Y’ format, and explain what it means in plain language: [insert data]”
2. Show absolute risk alongside relative risk: Relative changes can sound dramatic without a baseline.
Example: “This doubles your risk” → “Your risk goes from 1 in 1,000 to 2 in 1,000.”
Prompt: “Convert this relative risk into absolute risk and present them together in a way that adds context: [insert data]”
3. Provide a baseline or comparison point: A number on its own can be hard to judge as high or low.
Example: “There is a 5% chance of side effects with this drug treatment” → “About 5 out of 100 people who take this drug have side effects. Without treatment, the risk of these effects is about 3 out of 100.”
Prompt: “Add a brief comparison or baseline to help interpret this risk. If appropriate, compare to doing nothing or to a common reference point: [insert data]”
4. Align descriptive words with actual numbers: General terms like “rare” or “common” can be hard to interpret.
Example: “A rare side effect” → “A side effect that affects about 2 out of 100 people.”
Prompt: “Check whether words like ‘rare,’ ‘common,’ or ‘small risk’ match the numbers here. Suggest more accurate wording or add context if needed: [insert text]”
5. Keep formats and denominators consistent: Inconsistency in a document makes risk information harder to follow.
Example: “5% of patients…” in one section and “1 in 20 patients…” in another → revise to use the same format throughout.
Prompt: “Review this text and flag any inconsistencies in how risk is presented (percentages vs. frequencies, changing denominators, etc.): [insert text]”
6. Pair risk with clear action: People need to know what to do with the information.
Example: “This increases your risk of heart disease.” → “If you’re concerned about your risk, talk with your doctor about ways to lower it, such as diet, exercise, or medication.”
Prompt: “Based on this risk information, suggest a clear, appropriate action for the reader. Keep it aligned with the level of evidence: [insert text]”
Bonus: AI tools like ChatGPT can also help you create simple visual aids to explain risk. Here’s one I created to show the risk of breast cancer:
Here’s the prompt I used to create it:
I want to create a visual aid to help people understand how breast cancer risk changes if a woman has a close relative with the disease. Please use the statistics below from Breastcancer.org (https://www.breastcancer.org/facts-statistics) as the basis for the graphic. The audience is the general public looking for health information. Assume average numeracy and health literacy.
Suggest a type of visual aid that would work well with this data and ask me questions about what I’d like the presentation to be like.
Statistics: Approximately 1 in 8 women (13%) in the U.S. will develop invasive breast cancer at some point in their life.
A woman’s risk of breast cancer nearly doubles if she has a first-degree relative (for example, their mother, sister, or daughter) who has been diagnosed with breast cancer. Approximately 15% of women who get breast cancer have a family member diagnosed with it.