Founder used Anthropic’s Claude to manage cancer treatment data and navigate rare diagnosis

Connor Christou, founder of the AI-powered medical practice platform Keragon, used Anthropic’s Claude to manage and synthesize data during treatment for aggressive non-Hodgkin’s lymphoma, a rare diagnosis affecting roughly one in 420,000 people.

Christou, who had previously tracked over 100 biomarkers annually and maintained detailed logs via wearables and journaling, began feeding blood results, wearable outputs, scan data, and voice-transcribed symptom journals into Claude to cross-check oncologist recommendations and interpret ambiguous medical findings.

After receiving conflicting treatment recommendations from two oncologists—one advising a lighter chemotherapy regimen with about a 60% success rate and another recommending a more aggressive six-month inpatient protocol with roughly an 85% success rate—Christou gathered 12 expert opinions and ultimately chose the more intensive path. He described his approach as data-driven and borrowed from his prior military service, framing treatment as a series of finite cycles with measurable outcomes.

During treatment, Christou relied on his Whoop wearable to predict immune system lows and maintained a symptom journal, narrowing his focus to sleep, nutrition, and psychology. He fed all collected data into Claude, which he said helped him ask the right questions rather than replace doctors.

At the end of treatment, an ambiguous PET scan result led his oncologist to discuss additional therapies. After reviewing research indicating a roughly 60% false-positive rate on end-of-treatment PET scans for his specific lymphoma, Christou input three PET scans and an MRI into Claude, which identified a known but easily overlooked phenomenon: thymus gland reactivation in patients under 40, which can mimic active disease on imaging. The model estimated a roughly 90% probability of this explanation, and a fourth oncologist confirmed the diagnosis as thymus rebound, sparing Christou from unnecessary radiotherapy.

Christou’s experience reflects a broader trend: a public opinion poll released in March found that a third of American adults now use general-purpose chatbots for health information and advice, despite cautions from experts like Danielle Bitterman, clinical lead for data science and AI at Mass General Brigham, who has noted that such tools are frequently wrong and have not been thoroughly evaluated for personalized diagnoses.