Steve Brown
Steve Brown, Founder and Chief Executive Officer of CureWise (curewise.com), an artificial intelligence (AI)-driven patient advocacy app, describes his year-long quest to understand a series of symptoms that ultimately led to a diagnosis of light chain (AL) amyloidosis—a disease closely related to multiple myeloma—and how he used AI to help in his treatment decision-making.
In this interview with The ASCO Post, Mr. Brown expounds on his medical odyssey that combined consultations with both AI medical agents and their human counterparts; the trustworthiness of the answers he received from both; and his prediction that AI will enable the democratization of precision oncology care for all patients with cancer.
Discerning the Accuracy of Information From Both AI and Human Physicians
You describe how multiple physicians and extensive testing failed to identify an underlying plasma cell disorder, which was later diagnosed as AL amyloidosis. After AI agents you built to review your medical record found patterns suggesting a plasma cell disorder and bone marrow issues that your physicians had missed, these AI agents raised concerns after your physicians recommended the combination of daratumumab plus cyclophosphamide, bortezomib, and dexamethasone to treat the disease. The AI agents suggested instead discussing daratumumab plus the BCL2 inhibitor venetoclax with your oncologists, the treatment you are receiving today.
Given that the AI agents didn’t always agree on their proposed recommendations, and you weren’t using them to replace your oncologists’ clinical judgment, why did you trust AI’s suggestions enough to bring them into your care?
I learned early in this process that even excellent physicians can interpret the same data differently. My doctors were trying to make sense of a complex and unusual presentation, and their opinions didn’t always line up. That’s not a criticism. It’s a reflection of how much uncertainty exists in medicine, especially with rare plasma cell disorders.
AI had its own uncertainty as well. If you use AI expecting one perfect answer, that is the wrong expectation. I never saw AI as the final arbiter in my care. Its value is in detecting patterns. Different models may disagree on the details, but when they start to converge and cross-validate each other, that alignment becomes meaningful. In my situation, AI helped me to see where to focus my attention and which questions to bring to my oncologist.
Once I finally had the correct diagnosis, the value of asking informed questions became even more important. My initial treatment was a standard regimen for AL amyloidosis, but the treatment response plateaued soon after. The AI models drew my attention to the significance of the t(11;14) translocation in my type of AL amyloidosis and to emerging data showing the clinical benefit of daratumumab plus venetoclax in the treatment of this cancer.
I didn’t accept AI’s suggestion as a recommendation. I took it as a signal to get more expert opinions, which I did by consulting with hematologists at major cancer centers. After those conversations, my oncology team and I agreed to pursue off-label treatment with daratumumab plus venetoclax, which has halted disease progression. I’m now receiving the combination as maintenance therapy.
AI didn’t diagnose me, and it didn’t choose my treatment. It helped me understand the disease landscape I was in, frame better-informed questions, and participate more effectively in the decision-making process with my oncology team.
Establishing a New Paradigm for Investigating Cancer Therapies
As AI helps precision medicine become a reality for patients with cancer, potentially individualizing every stage of care from diagnosis to treatment selection, will randomized controlled clinical trials become unnecessary or even obsolete?
We know cancer isn’t just one disease, but a collection of hundreds of molecularly distinct conditions. AI doesn’t remove the need for evidence-based research but exposes the limits of the clinical trial system that produces it. The more precisely we define a tumor and the more targeted the therapies become, the smaller each patient subgroup will get. At a certain point, these groups will become so small that traditional randomized controlled trials will not be feasible—effectively, each patient becomes an N-of-1.
I saw this situation firsthand in my own care. One of the specialists I consulted had planned a clinical trial of daratumumab plus venetoclax for patients with t(11;14) AL amyloidosis. The biology made sense, and both pharmaceutical companies supported the study, but the trial was cancelled because he couldn’t find enough patients who matched the molecular profile.
That’s the reality of precision oncology. The science keeps getting more exact, and the study cohorts keep getting smaller, which is why we need a new paradigm for evidence. Randomized clinical trials will still matter when they are possible, but precision medicine requires additional tools, from adaptive study designs and real-world evidence to AI systems capable of learning from very small molecular subgroups.
We’re moving toward a model of research of continuous learning rather than one-time validation, because biology is now more individualized than traditional methods were designed to study.
Helping Patients Make Better Informed Care Decisions
You are a seasoned AI and technology expert with the background and resources that allowed you access to cancer specialists and to use AI to help you understand your diagnosis and treatment options. Many patients diagnosed with cancer do not have those advantages. Getting a cancer diagnosis is already a daunting experience. Isn’t it a lot to ask patients to take on the added responsibility of using AI to help in their treatment decision-making process, even if it is in consultation with their oncology team?
Getting a cancer diagnosis is already a lot for anyone to handle. Some patients are comfortable saying to their oncologist, “Just tell me what to do.” Others may feel they need to take a more active role in their care to get the best outcome, and that usually means becoming a self-advocate or asking a family member to help with their advocacy.
We created CureWise for that second group. The goal is to help patients understand their specific cancer well enough to feel some level of control in the decisions they face throughout treatment and survivorship. The technology is not meant to add work or stress to an already difficult experience. It is meant to give patients clear, accurate information, so they can make informed choices with their oncology team.
If you use AI expecting one perfect answer, that is the wrong expectation…. AI helped me to see where to focus my attention and which questions to bring to my oncologist.— STEVE BROWN
Tweet this quote
In my case, for example, asking AI a simple question about whether it mattered if I took venetoclax in the morning or at night led to a practical detail I had not been told by my oncologist. AI said the time of day I took the drug didn’t matter, but it should be taken with about 40 g of fat, which affects how the medication is absorbed. There are many other lifestyle factors that can influence the effectiveness of treatment, and AI can help bring them to the surface.
CureWise does not diagnose patients, and it does not tell them which treatments to pursue. Their oncologist will guide that decision. What CureWise does is help patients understand their disease, know which questions may be important to ask their oncology team, and find appropriate clinical trials and emerging therapies that may be relevant to their care.
Advancing Patient Cures Through Precision Medicine
Can patients access CureWise for free?
We offer a free level that allows patients to use AI for an initial assessment of their cancer type and possible treatment options, along with access to a clinical trials finder. We are building CureWise as a full management tool to support patients throughout the course of their cancer journey, so they have the tools they need to understand their diagnosis and treatment possibilities and have more productive conversations with their oncologist. Some tools will remain free, while others require a subscription, because developing this type of technology is expensive.
Our mission is to advance cures through precision medicine. We are focused on patient experience and providing tools that help them become genuine partners in their care, while also supporting oncologists in the very difficult work they do. There is a lot of stress in the health-care system. We are designing tools to help make the experience better for everyone.
Mitigating Disparities in Cancer Care
Will integrating AI and large language models into clinical care help decrease disparities in cancer care; and are you concerned that the technology could make existing disparities worse by introducing algorithmic biases into the system?
I think AI can help reduce some disparities in cancer care, but it cannot solve all the problems. Lack of health insurance or access to high-quality medical services is a societal problem that technology alone cannot fix. What AI can do is help lower some of the practical and informational barriers that many patients face.
For patients living in rural areas or far from major cancer centers, AI can make virtual access to oncology resources easier and reduce the need for long-distance travel for clinical care. In addition, by helping clinicians interpret genomics and patients’ clinical histories more precisely, AI can support more personalized treatment decisions for every patient—not just those treated at large academic cancer centers.
It is also true that AI systems can reflect the biases in the data used to train them. Mitigating those biases requires active work, including data collection across diverse patient populations, auditing frameworks to ensure accountability and trust, and careful evaluation of AI models. This is essential if AI is going to serve all patients equitably.
However, we should recognize that many disparities in precision oncology are the result of lack of information. Patients often do not know to ask about appropriate tests, what treatments may be effective against their specific cancer, and how to find clinical trials. AI can help bridge those gaps by providing patients with practical knowledge and making them more active participants in their shared decision-making.
Patients who learn how to use AI may find it easier to navigate an increasingly complex health-care system, become more active participants in their care, and regain some sense of control.
DISCLOSURE: Mr. Brown is the Founder and Chief Executive Officer of CureWise.
