Conno Christou’s approach to AI cancer treatment decisions began not in a hospital, but in a moment of accidental discovery: a swollen arm after a workout that turned out to be blood clots, which turned out to be covering something far worse.
Pre-operative scans revealed an 11-by-11-by-8 centimetre mass behind his sternum. A biopsy confirmed an aggressive, fast-growing form of non-Hodgkin’s lymphoma, a rare diagnosis affecting roughly one in 420,000 people, caused by a random genetic mutation with no connection to lifestyle, diet, or stress. Christou was 35. His annual bloodwork, checked just months earlier across nearly 100 biomarkers, had come back clean. ‘It was the best I’d had in years,’ he says.
The tumour had existed for about three months. In three more weeks, it would have reached stage four.
Two World-Class Doctors, Opposite Advice
What followed demonstrated both the limits of specialist medicine and what a motivated, data-literate patient can do within those limits. Christou’s first oncologist recommended the lighter of two available chemotherapy regimens. He booked his first infusion three days out, then sought a second opinion the night before. That second doctor recommended the harder path: continuous in-hospital infusion, cycling every three weeks across six months. The lighter treatment carried roughly a 60% success rate for his presentation; the aggressive regimen brought that to around 85%.
He didn’t simply go with doctor number two. Over the next two days, he gathered 12 opinions in total, reaching out to hematologists and oncologists in the US and abroad. Eleven to one voted for the harder regimen. He took it.
‘As founders, we hold the wheel,’ he says. ‘You hear many things. You don’t have to follow the first advice.’
Christou treated the six cycles of chemotherapy the way he approaches company-building: a marathon of sprints, each with a finite arc and a clear set of data points. He wore his Whoop band throughout treatment and found it accurate at predicting the days his immune system would bottom out, sometimes flagging them before symptoms arrived. He kept a symptom journal via voice transcription. He narrowed his focus to three variables: sleep, nutrition, and psychology. ‘It moves the needle more than anything,’ he says. ‘I never asked “why me”, not once. That question has no useful answer.’
AI Cancer Treatment Decisions at the Critical Moment
He fed his blood results, scan data, wearable output, and journal entries into Claude. A Mass General Brigham researcher, Danielle Bitterman, a faculty member in the institution’s AI in Medicine programme and a radiation oncology physician at Brigham and Women’s Hospital, has warned that general-purpose chatbots are frequently wrong and ‘have not been thoroughly evaluated’ for personalised diagnoses. Christou doesn’t disagree with the caution. ‘It didn’t replace the doctors,’ he says, but it ‘helped me ask the right questions.’
For a condition so rare that an oncologist might see it once a year, access to a model trained on the full body of medical literature was categorically different from a Google search.
The model proved most consequential at the end of treatment. His final PET scan came back ambiguous. His oncologist began discussing radiotherapy near his heart and lungs. Christou read that, for his specific lymphoma, the false-positive rate on end-of-treatment PET scans runs at around 60%, a figure supported by peer-reviewed research on lymphoma imaging examining the limited predictive value of post-therapy scans. ‘It’s 2026. Sixty percent,’ he says.
He fed all three of his PET scans and his MRI into Claude. The model flagged a known but easily missed phenomenon: in patients under 40 recovering from this type of lymphoma, the thymus gland can reactivate after chemotherapy, producing imaging appearances that resemble active disease. Given his age and scan characteristics, the model put the probability of that explanation at roughly 90%. Research published in Pediatric Blood and Cancer has found thymic rebound rates of between 44% and 67.7% in younger lymphoma patients after chemotherapy, though that research focuses on paediatric cases. Christou sought three more opinions. The fourth doctor confirmed it: thymus rebound, no active disease, no radiotherapy needed.
He was clear.
Building the Tool He Wish He’d Had
There is an irony in all of this. Christou built Keragon, his Boston-based AI platform for healthcare practice automation, before any of this happened. The company integrates with more than 300 healthcare and business applications, helping provider teams handle scheduling, patient intake, eligibility checks, and data sharing without writing code. CIOReview named it a Top 50 Healthcare Solutions Company for 2024. Going through the system as a patient gave him a different vantage point: nurses and doctors buried in administrative tasks, chemotherapy protocols standardised across wildly different patients, side effects managed by cascading chains of additional drugs. ‘I’m certain we will look back at this era of treatment and cringe,’ he says.
He takes Sundays off now, mostly. A VC friend told him something years ago that he replayed throughout treatment: be happy now. He says it’s among the hardest things to do, and the one thing he has finally come to appreciate.
‘It’s not happening in 10 years,’ he says of what AI can already do for patients willing to use it. ‘It’s happening today.’ Whether the medical establishment catches up before the next ambiguous scan lands on someone else’s desk is the question worth watching.
