This is how a patient with metastatic cancer uses AI to understand her disease: what is the best method to talk about health with chatbots?

In 2021, Miriam González, a 35-year-old woman from Murcia, went to the doctor for breast bleeding. They told her not to worry, that everything was normal. But in 2024, she was diagnosed with breast cancer and, shortly after, found out it was stage four metastatic. “At first, I thought the diagnosis was an immediate death sentence, that I had days or weeks left,” González explains in an exchange of messages with EL PAÍS. But it wasn’t like that, she had some leeway: “I started hearing about chronicity and quality of life, and I saw that the outlook today is different. That mental transition, from ‘I’m dying now’ to ‘I’m going to live with this,’ was tough and I needed to understand what ground I was on,” she explains.

Read more Express trial for the death of Iliass Tahiri, the 18-year-old suffocated seven years ago at the Oria juvenile center (Almería)

To understand that ground, she turned to Perplexity, an artificial intelligence (AI) search engine. Then “my engineer side emerged,” says González, “to break down the problem.” Miriam’s case is special. Her tumor had neuroendocrine differentiation: “It’s such a rare subtype that standard clinical guidelines simply don’t cover it,” she says. AI helped her understand, to “organize that complexity and convert an abstract diagnosis into concrete decisions,” she adds.

Millions of people already use AI as a jargon translator, medication consultant, or directly as a doctor. But it is necessary to differentiate between these uses, warns Mark Succi, director of healthcare innovation at Mass General Brigham, a hospital network in Boston, and associate professor at Harvard: “AI seems more useful in the later and more confined stages of diagnosis, narrowing the focus towards an answer once the case is already structured, and less useful in generating an initial diagnostic framework aware of uncertainty.”

A study published last week, which analyzed five of the most popular models, such as Gemini and ChatGPT, showed that half of the health information provided lacks scientific rigor, an inaccuracy that puts patient safety at risk.

However, a new survey in the US reveals that one in four Americans use chatbots for health questions. The reasons they give are because they want “quick answers” or “additional information.” There are also people who want to research on their own before or after seeing a doctor. But there is a significant group that uses it instead of a doctor, especially people with low incomes (in the US, healthcare is not public): 32% of users with incomes below $24,000 annually turned to AI because they could not afford a medical visit.

González’s case is different. She prepared a more personal journey and did it together with her doctors: “I’ve been lucky enough to find a team that truly includes me. They listen to me, read the evidence I provide, question with me and not against me,” she says. The engineer assures that, without AI, she probably would have intuited the rarity of her tumor, but would not have been able to access all “the data, trials, case series, or technical language to turn it into a proposal that oncologists could take seriously,” she explains.

For this, she turned to an AI specialist she already knew, Javi López, also from Murcia and co-founder of Magnific. “There came a point when I needed someone who could handle more advanced tools and take what I was finding to another level. That’s when Javi came in,” she says. González believed that López would be able to give her research more momentum. Both shared their case on X, which went viral.

With ChatGPT printed

Doctors see it from the other side. Oriol Mirallas, a medical oncologist in the Phase 1 Experimental Therapies unit at MD Anderson Cancer Center in Houston (University of Texas), understands that it is inevitable but delicate: “We are seeing more and more people coming with ChatGPT or clinicaltrials(.)gov [reference database for clinical trials] printed out. Here in the US, it’s even more common. It’s reasonable for patients to seek help, and AI can provide it, but with the help of an expert. If it helps the patient understand the pathology and diagnosis, that’s fantastic. But finding feasible and optimal treatments, in a field that changes daily, is complicated,” he says.

Read more Ten million memories

Ultimately, the distance between both worlds, engineers and doctors, is not that great. Both believe that AI is an inevitable progress, that it will be used more, but that a human must always have the final say. The problem is the weight of each of these factors. “It’s exciting that we have more tools to empower and educate patients,” says Arya Rao, a researcher at Harvard University. “I am optimistic about AI’s potential to personalize patient education, but doctors are responsible for clinical AI. Instead of discouraging patients from using these tools, doctors should open the conversation: ask them what they have searched for, what AI has told them, and what questions they have,” she adds.

AI has its method

The complexity of AI’s medical use lies in the sophistication it can already achieve today. Javi López’s response for González’s case was tremendously refined: first, he used the most advanced systems, ChatGPT Pro+ Extended and Claude Opus 4.6 MAX. “These models, in their most powerful versions, cost about 200 euros a month,” he says. Second, he converted González’s entire medical history into a text document, to have all the information together.

Then he created a prompt also made with AI of almost 2,000 words where he told the AI that it was a “multidisciplinary tumor committee composed of the best specialists in the world.” Once he had the response from one model, he passed it to the other to look for flaws: “This kind of ‘adversarial model’ has always worked. It’s the same as with humans: two research teams in parallel sharing their discoveries are usually more productive than one,” he adds.

Would this system work for other types of diseases or if doctors used it on their own? For López, it’s obvious that it would: “In the near future, I hope that everyone’s medical history will not only be digitized, but also ‘chewed up’ so that it can be consumed by AI, so that any doctor can make a consultation with all your history and years of results at their fingertips.”

It’s not an easy path, but it’s already being explored by major Silicon Valley companies. In January, OpenAI released ChatGPT Health, where users can upload their medical records. But today there is still a difference, depending on who is in charge of the AI: “I am aware that not everyone can do what I do,” says González. “Having time to research, knowing how to read scientific literature even with help, building an international network of contacts while undergoing treatment. That’s why I think it’s important to say it out loud: not to serve as a model, but as an argument in favor of these tools and this type of support being available to anyone,” she adds.

On social media, González’s case has been compared to that of Sid Sijbrandij, co-founder of GitLab, a software collaboration platform. He was diagnosed with osteosarcoma with no available trials. He used AI to analyze 25 terabytes of data from his tumor, identified the overexpression of a protein, traveled to Germany to receive a therapy targeting that marker, and today his cancer is undetectable. “The logic is the same,” says González. “When guidelines don’t cover your case, AI can help you find a path that does. But it’s important to be honest: Sid had access to technology and resources that most patients, myself included, do not have. If there’s one thing I advocate for, it’s that this way of navigating the disease shouldn’t depend on what one can afford,” she adds.

These are unique cases. González’s example is closer to reality, but also special. They don’t serve as a model, but they are a hint that everything has already changed, including in medicine, says Mark Succi: “Doctors should treat this as a permanent part of modern healthcare and respond without falling into the error of dismissing new tools. The best response is to explain in which cases they can be useful and in which they cannot. These systems can sound confident even when their reasoning is weak, especially in complex cases. That’s why doctors should help patients use AI results as a starting point, not as a diagnostic conclusion,” he says.

Read more Why does the Bernabéu boo?