Transforming cancer care with generative AI
Can cutting-edge AI help humanity tackle devasting diseases like cancer?
Rakuten AI Optimism Conference, the company’s flagship annual business conference, offered a glimpse of that future with the panel “Accelerating Cancer Research with AI to Change Lives.”
Rakuten’s Taku Okoshi and Yusuke Kaji joined Dr. Toshihiko Doi, Director of Japan’s National Cancer Center Hospital East, to explore how generative AI can lead to better patient care and medical outcomes.
The conversation tackled both the promise and potential pitfalls of bringing large-language-model (LLM) technology to the front lines of cancer care—and showcased a working prototype already in the hands of researchers.
Why generative AI matters in healthcare
As healthcare systems face provider shortages, soaring data complexity, and rising patient expectations, generative AI offers healthcare providers the ability to:
- Bridge communication gaps between patients and medical professionals by understanding emotional, natural language.
- Summarize and organize dense clinical information so clinicians can make faster, better-informed decisions.
- Help patients articulate symptoms even when they’re unsure what questions to ask.
- Relieve operational pressures through automated triage, documentation, and follow-ups.
- Enhance continuity of care by remembering and linking context across visits and specialties.
Utilized responsibly, these capabilities can make care more empathetic, scalable and accessible—without replacing the human touch.
Inside Rakuten’s “Proactive Medical AI” prototype
During the session, Kaji unveiled a live demo of Rakuten’s Proactive Medical AI, which showcased the following process:
- A patient describes symptoms in everyday Japanese.
- The AI conducts a conversational check against a clinician-curated checklist, filling in missing details with follow-up questions.
- It then generates a concise, structured summary—complete with urgency flags and suggested next steps—for the attending physician.
Unlike scripted chatbots, the model adapts to emotion and nuance, helping patients feel heard while lightening the clinician’s workload.
Guardrails: Building for safety, not hype
All speakers underscored the importance of responsible deployment of generative AI technology in healthcare. This includes:
- Human-in-the-loop review for every critical output
- Transparent confidence scores that highlight uncertainty
- Continuous validation against real-world clinical data
- Clear escalation paths whenever the AI reaches its knowledge limits
As Dr. Doi reminded the audience, “In general, even experts would make mistakes; what matters is designing systems that catch them before patients are affected.”
What this means for patients and clinicians
For patients, this means:
- 24/7, judgment-free assistance that speaks their language
- Faster answers during moments of anxiety or confusion
- A documented voice in their own medical journey
For clinicians:
- Condensed, prioritized patient histories before the exam even begins
- Fewer administrative hours spent on charting and follow-up calls
- Decision support that highlights—but never overrides—clinical judgment
For the care team as a whole, these systems can offer a shared, living context that travels with the patient, reducing repeat questions and improving continuity.
Looking ahead
The Rakuten–National Cancer Center team have begun pilot trials to measure the AI system’s impact on consultation time, diagnostic accuracy and patient satisfaction. Findings will feed directly into improving AI agent experiences.
“AI isn’t here to replace people,” Okoshi concluded. “It’s here to make every conversation more complete, every diagnosis more informed and every patient feel more understood.”
Disclaimer: The capabilities described are under research and are not yet cleared for clinical use. Medical decisions should always be made by qualified professionals.
