Tech • AI • Robotics
This doctor uses a robot and his own AI for his patients
5/10TL;DR
Julien Vellementand, a radiation oncologist in Montpellier, is using robotics and artificial intelligence to make cancer care more humane, especially for children undergoing radiotherapy.
KEY POINTS
A physician at the crossroads of medicine and technology
Julien Vellementand works in radiation oncology at the Institut du cancer de Montpellier and at the Onkodoc radiotherapy center. He chose the specialty partly because radiotherapy has long relied on heavy computing, imaging, data processing and highly technical machines. In his view, the field naturally became one of the first medical areas where AI could be integrated into routine practice.
A lifelong technology enthusiast
Vellementand traces his interest in technology back to early childhood, when he received a computer at age 6 and began taking machines apart. He describes that trajectory not as a sudden shift from medicine to technology, but as a continuous path that led him toward a specialty combining patient care with advanced equipment. He rejects the idea that technical rigor requires sterile or impersonal practice, arguing that medicine can be serious without being austere.
A treatment setting that can be traumatic for children
Pediatric radiotherapy places children in an especially difficult situation. Sessions typically last about 10 minutes a day, Monday to Friday, over one to two months, inside a shielded treatment room with thick concrete and lead walls. Unlike surgery or chemotherapy, where parents can often remain nearby, radiotherapy requires children to stay alone in the room while adults monitor from outside because of radiation exposure.
A child’s case triggered the search for another solution
The turning point came during the care of a 5-year-old girl who underwent a particularly hard treatment lasting one hour a day for two months. Despite every attempt to comfort her, including family voices over the intercom, she cried through every session. The experience convinced Vellementand that existing tools were not enough for some children and that a new form of support was needed.
From an idea in Japan to the first robot of its kind in care
While traveling in Japan, Vellementand saw a small robot hand him a towel with a smile after he came in from the rain. The encounter sparked the idea that a robot could enter spaces where people could not, including radiotherapy rooms, and still provide emotional reassurance. After abandoning plans to build one alone and finding academic development far too costly at an estimated 5 million euros and five years of work, he contacted the French team behind Miroki, later also developed as Miroka.
A world-first use of Miroki in pediatric radiotherapy
Vellementand says he was the first in the world to integrate Miroki into this medical use. Originally designed for social logistics rather than direct caregiving, the robot relies on nonverbal communication, including moving ears, facial expressions and a friendly visual presence. That design made it suitable for supporting children in a frightening environment without replacing medical staff or parents.
Montpellier has made the robot part of standard care
The robot is now a standard element of pediatric treatment in Montpellier for children receiving radiotherapy. Vellementand’s goal is to extend the model internationally, noting that around 400,000 children develop cancer each year worldwide and that roughly one third will need radiotherapy. He is now in discussions with centers in France, the United States, Quebec and Japan, and is presenting the project at major professional meetings including one in Boston.
Administrative skepticism gave way to support
The project initially faced resistance from hospital leadership, especially before generative AI became widely discussed. Approval required a detailed roadmap, evidence of usefulness and funding for the robot itself. Financial backing ultimately came through fundraising with help from a child’s mother, while the collaboration with the robot’s manufacturer evolved into a close partnership rather than a simple supplier relationship.
AI is seen as an accelerator, not a substitute for clinicians
Vellementand argues that much of today’s medical AI remains overmarketed, but he expects it to become genuinely transformative within about 10 years as it converges with biology, genomics, robotics and other tools. He points to cancer classification as an example: where clinicians once relied on broad categories, they now use increasingly detailed biological subtypes, and future personalization will exceed what a human brain can process alone. In one difficult case, AI helped identify a very rare diagnosis far faster than clinicians likely could have unaided.
Education and training remain the weak link
He believes medical education must be redesigned for an era in which raw knowledge retrieval is no longer the physician’s main advantage. Training a doctor over 12 years without fully integrating AI, he argues, risks preparing students for a world that will already be obsolete by the time they graduate. He is also critical of fragmented hospital systems and says progress will remain limited until health tools become interoperable.
CONCLUSION
Vellementand’s work shows how robotics and AI can be applied not just to optimize care, but to reduce suffering in some of medicine’s most difficult moments. The broader challenge now is whether health systems can scale those gains fast enough through funding, interoperability and training.
Explain this