Improving patient understanding in metastatic colorectal cancer using chatgpt like tool for communication

Effective communication between physicians and patients is paramount, especially when discussing complex conditions like cancer. Studies have shown that clear patient-doctor communication improves treatment adherence, patient satisfaction, and overall outcomes1. However, explaining intricate biological mechanisms and treatment modalities to patients with varying levels of health literacy can be challenging.

Metastatic colorectal cancer (mCRC) involves complex processes such as cellular mutations, metastatic spread, and multifaceted treatment options including chemotherapy, targeted therapy, and immunotherapy2. Simplifying these concepts without losing essential information is crucial for informed consent and patient empowerment.

Artificial intelligence, particularly language models like ChatGPT developed by OpenAI, offers innovative solutions to this communication barrier. By generating patient-friendly analogies, tools like ChatGPT can assist oncologists in conveying complex medical information in an accessible manner3. This article aims to guide healthcare professionals in crafting effective prompts to utilize ChatGPT for enhancing patient understanding of mCRC.

Using tools like ChatGPT to create clear analogies

To maximize the utility of tools like ChatGPT, it’s essential to formulate precise prompts that consider the target audience’s comprehension level4. Below are techniques and examples to help oncologists generate meaningful analogies for explaining mCRC concepts.

Explaining complex mechanisms in cell reproduction

Technique: Request analogies that relate cell reproduction to everyday concepts, defining the target audience as patients with lower education levels.
Prompt example 1: “Explain how cancer cells grow and divide in metastatic colorectal cancer using an analogy suitable for a patient with a basic education level.”

AI response: “Imagine your body’s cells as a team of workers building and maintaining a city. Each worker knows their job and follows the rules. Some workers in the colon district ignore the rules and start building nonstop, creating chaotic structures that harm the city. These unruly workers even travel to other parts of the city, causing more trouble elsewhere.”

Prompt example 2: “Describe the process of normal vs. cancerous cell division in metastatic colorectal cancer using a simple analogy for a patient who may not have a medical background.”

AI response: “Think of cell division like a photocopy machine making copies of a page. Normally, the machine makes clear copies and stops when done. In colorectal cancer, the machine malfunctions, making endless copies with errors, and doesn’t stop, leading to a messy pile of unusable papers that clutter the office.”

Prompt example 3: “Use an everyday analogy to help a patient understand how cancer cells spread in metastatic colorectal cancer.”

AI response: “Picture weeds growing in a garden. If not removed, they not only overtake the flowers but their seeds can scatter and grow new weeds in other parts of the garden. Similarly, colorectal cancer cells can spread from the colon to other parts of the body if not controlled.”

Explaining metastatic colorectal cancer as a complex condition

Technique: Use familiar scenarios to illustrate the complexity and seriousness of mCRC, ensuring the analogy resonates with the patient’s experiences.

Prompt example 1: “Explain metastatic colorectal cancer to a patient by comparing it to a situation they might encounter in daily life.”

AI response: “Metastatic colorectal cancer is like a small fire starting in one room of a house. If not quickly contained, the fire can spread through the halls to other rooms, causing more damage. Early detection and action are crucial to prevent the fire from engulfing the entire house.”

Prompt example 2: “Help a patient understand metastatic colorectal cancer by using an analogy related to a common community issue.”

AI response: “Think of it like a rumor starting in a small town. It begins in one neighborhood but quickly spreads to others, causing widespread confusion. Similarly, cancer that starts in the colon can spread to other parts of the body, affecting overall health.”

Prompt example 3: “Describe the seriousness of metastatic colorectal cancer using an analogy appropriate for someone without medical knowledge.”

AI response: “Imagine a crack in a dam. If it’s not repaired, the crack widens, and water starts leaking through, eventually flooding areas downstream. Metastatic colorectal cancer starts in one spot but can expand and impact other parts of the body if not treated.”

Explaining main categories of treatments

Technique: Simplify treatment modalities by likening them to solutions in familiar problem-solving scenarios.

Prompt example 1: “Use an analogy to explain how chemotherapy works in treating metastatic colorectal cancer to a patient with limited medical understanding.”

AI response: “Chemotherapy is like using a weed killer in a garden overrun by weeds. It aims to eliminate the fast-growing weeds (cancer cells) to allow the healthy plants (normal cells) to thrive.”

Prompt example 2: “Explain targeted therapy for metastatic colorectal cancer with a simple analogy suitable for a layperson.”

AI response: “Targeted therapy is like a locksmith making a key that fits only a specific lock. It targets unique features of cancer cells, leaving most healthy cells alone, much like opening only the door that needs to be opened without disturbing others.”

Prompt example 3: “Describe immunotherapy for metastatic colorectal cancer using an everyday comparison for better patient understanding.”

AI response: “Immunotherapy is like training a guard dog to recognize and chase away intruders. It helps your immune system better identify cancer cells (the intruders) and attack them more effectively.”

Techniques to Improve Accuracy

1. Specify the Audience: Clearly define the patient’s education level and background in the prompt.
   • Example: “Explain X to a patient with a high school education.”
2. Request Analogies: Ask for explanations using analogies related to common experiences.
   • Example: “Use a household analogy to describe how Y works.”
3. Limit jargon: Instruct to avoid medical terminology unless it’s explained.
   • Example: “Explain Z in simple terms without medical jargon.”

Limitations to the use of LLMs in patient education / communication

While tools like ChatGPT are a powerful tool for generating patient-friendly explanations, it’s essential to review and verify the content it produces. AI language models can sometimes provide incorrect information or misleading analogies5. Given that the GPT chatbot has difficulty understanding the full context of a question, especially in the case of complex specialized questions, always ensure that the explanations are accurate, appropriate, and align with medical facts before sharing them with patients.

Conclusion

Enhancing patient understanding in metastatic colorectal cancer is crucial for effective treatment and care. By leveraging tools like ChatGPT to create clear and relatable analogies, oncologists can bridge the communication gap caused by the complexity of medical concepts. However, it’s imperative to critically assess the AI-generated content for accuracy and suitability. Combining professional expertise with innovative technology can significantly improve patient education and outcomes.

This article was written with the assistance of generative AI technology and reviewed for accuracy.

References

1. Holden CE, Wheelwright S, Harle A, et al. The role of health literacy in cancer care: A mixed studies systematic review. PLoS One. 2021 Nov 12;16(11):e0259815.
2. Shin AE, Giancotti FG, Rustgi AK. Metastatic colorectal cancer: mechanisms and emerging therapeutics. Trends Pharmacol Sci. 2023 Apr;44(4):222-236. doi: 10.1016/j.tips.2023.01.003. Epub 2023 Feb 23.
3. Cè M, Chiarpenello V, Bubba A, et al. Exploring the Role of ChatGPT in Oncology: Providing Information and Support for Cancer Patients. BioMedInformatics. 2024; 4(2):877-888.
4. Tips for generative AI large language model (LLM) prompt patterns, RedHat, last accessed 12/12/2024, https://www.redhat.com/en/blog/tips-for-gen-ai-prompts
5. How Can We Counteract Generative AI’s Hallucinations?, Harvard Business School, last accessed December 1^st, 2024 https://d3.harvard.edu/how-can-we-counteract-generative-ais-hallucinations/