A year ago, I was confronting a series of symptoms—including rapid weight loss, abdominal distress, fatigue, and heart issues—that I couldn’t explain. I was just 60 years old and had been in good health, but now I sensed that something was seriously wrong. I made appointments with my primary care physician, cardiologist, and a gastroenterologist, and asked for every test my insurance company would approve—about 100 in all, including full body scans, a colonoscopy, endoscopy, and cardiac function tests.
The verdict? Except for a couple of noncancerous polyps the gastroenterologist removed and some mild gastritis, the specialists could find nothing to explain my symptoms, and wrote off the tests as a fishing expedition. My cardiologist even suggested that maybe I was just stressed or depressed. Then on January 7, 2025, just a day after I got the final test results, the Palisades fire destroyed our house in Malibu, California, and my wife and I relocated to Palm Desert, California, and stayed with friends until we could figure out a permanent living situation.
Two weeks later, after dining out on a Friday night, I experienced unrelenting and severe abdominal pain and vomiting that continued over the weekend, finally sending me to the emergency room. Now, I was in a completely different medical environment from the one I had been in just a few weeks before. These physicians, unaware of my previous medical history, examined me with fresh eyes, and found that I had very enlarged abdominal lymph nodes.
“It’s cancer until proven otherwise,” said one of the physicians, and he admitted me to the hospital.
Building AI Agents to Sort Through Complex Health Data and Provide Answers
Although a biopsy of one of the lymph nodes didn’t find cancer, it did show amyloid proteins. That finding led to a bone marrow biopsy, which confirmed a rare blood cancer related to multiple myeloma. And while the cancer was detected early, it had already started to affect my kidneys, gut, and heart.
Lying in the hospital bed for 9 days, awaiting more test results and absorbing this new information, I couldn’t help but wonder why my previous medical team and all of the tests I underwent hadn’t picked up that I have a life-threatening blood cancer, and whether artificial intelligence (AI) would have caught the cancer sooner.
I was not unfamiliar with this technology. Two decades before my diagnosis, I helped pioneer remote patient monitoring and chronic disease management systems, and, more recently, I had created educational AI agents that could navigate complex questions. I relied on that experience now to determine my next steps.
“As genomics and molecular profiling become more precise, we’re learning that every cancer is biologically unique, which creates the next challenge in medicine.”— STEVE BROWN
Tweet this quote
I developed a medical AI agent I called “Haley” to analyze my medical record with a carefully designed medical context using underlying foundation models from OpenAI, Google, Anthropic, and xAI, and uploaded all of my data from MyChart, including notes from my physicians as well as laboratory and imaging test results. Within seconds, Haley conveyed that a concerning pattern of mild anemia, elevated ferritin, and low immunoglobulin levels could be explained by a plasma cell disorder and bone marrow issues, none of which had been previously suggested by earlier physicians. Haley had digested the same data but came to a different conclusion.
I repeated the experiment with more AI agents that I designed with a diverse set of perspectives. Even though the various AI agents didn’t agree on an exact diagnosis, they all suggested getting a serum free-light chain test to measure the levels of free kappa and lambda immunoglobulin light chains in my blood to determine if I might have a plasma cell disorder like multiple myeloma. The test confirmed that I, in fact, have light chain (AL) amyloidosis, a plasma cell disorder related to multiple myeloma. If I had gotten that test a year earlier, the cancer would have been diagnosed sooner, but I didn’t know to ask for that test, and none of my physicians had suggested the test.
Understanding Cancer to Be Molecularly Unique in Every Patient
Inspired by the depth of medical reasoning possible with AI, I then created my own virtual multidisciplinary medical team consisting of an oncologist, gastroenterologist, hematologist, cardiologist, emergency room physician, and a primary care physician, among others, all trained to process information like their human counterparts, for suggestions on the next vital step in my care: treatment options.
My human oncologist recommended the standard induction protocol for newly diagnosed AL amyloidosis, which included the targeted therapy daratumumab plus a chemotherapy regimen of cyclophosphamide, bortezomib, and dexamethasone (Dara-CyBorD). As my AI agents reviewed my bone marrow biopsy genomics, however, I learned that my cancer cells harbored the t(11;14) translocation.
While t(11;14) AL amyloidosis isn’t inherently indicative of an aggressive phenotype, it can be less responsive to the standard of care; it can also be exquisitely sensitive to venetoclax, a BCL-2 inhibitor. Although venetoclax is not United States Food and Drug Administration–approved in the treatment of this cancer, it has demonstrated efficacy in patients with t(11;14) AL amyloidosis, and is used off-label by some of the major cancer centers.
Sure enough, the Dara-CyBorD treatment plateaued after just a month. My AI models suggested that I talk with my oncologist about switching to daratumumab plus venetoclax right away rather than waiting for the Dara-CyBorD combination to fail, since my organs would continue to be damaged every day if I did not get the disease under control.
Reluctant to rely solely on what AI was suggesting, I consulted with hematologists from several leading cancer centers, including the principal investigator proposing the first phase II clinical trial studying the daratumumab-venetoclax combination therapy.
I learned that early clinical trials of daratumumab and venetoclax haven’t been conducted because not enough patients could be found to enroll in the studies. Indeed, the trial I had found had already been cancelled because not enough eligible patients could be enrolled in the study. This is the challenge with precision oncology at the present time. As genomics and molecular profiling become more precise, we’re learning that every cancer is biologically unique, which creates the next challenge in medicine. Traditional randomized trials cannot be conducted when each patient is essentially an N-of-1. The process can be tested, but not the treatment itself.
Encouraging Patients to Become More Proactive in Their Care
The combination of coaching from my AI tumor board and the clinical judgement of my human oncology team helped me reshape my treatment plan, and I’m now receiving daratumumab plus venetoclax off-label. I have had a remarkable complete response to the therapy, and I will continue to receive this drug combination as maintenance therapy for another year.
This experience has given me hope that perhaps AI can enable greater shared decision-making in oncology care, to allow patients and oncologists to work more collaboratively and encourage patients to become more proactive in their care.
Originally, I put together the various AI models to help me in my cancer journey. Now, I want to do the same for other patients with the disease. I’ve launched an AI-driven medical advocacy app called CureWise (curewise.com), which encapsulates everything I’ve learned as both a technology pioneer and a patient with cancer.
The goal for CureWise is to provide patients with a comprehensive resource for clinical trials and a streamlined management tool to help them achieve personalized insights tailored to their specific cancer type, so they can have more productive conversations with their oncologists. I believe that patients who become educated in their disease are going to be more effective partners in their care.
AI doesn’t replace anyone; it augments everyone. I went from being a victim of cancer to a proactive participant at every level of my care. That’s my hope for all patients with the disease.
(Editor’s Note: For additional information on Steve Brown’s experience using AI in conjunction with his human oncology team, see “Using AI to Ensure That All Patients With Cancer Have Access to Precision Oncology Care,” published in the January 25, 2026, issue of The ASCO Post.)
Steve Brown, 60, is the Founder and CEO of CureWise. He is a veteran entrepreneur and technologist focused on advancing patient-centered innovation in health care, and previously served as Chief AI Officer of Abundance360.
Editor’s Note: Columns in the Patient’s Corner are based solely on information The ASCO Post received from patients and should be considered anecdotal.
