Health

Direava’s Surgical AI Model Wins CEATEC Award 2026 Innovation Prize

· 6 min read · Health

Direava’s Surgical AI Model Wins CEATEC Award 2026 Innovation Prize

A Japanese AI model built to understand what is happening during an operation has won one of this year’s CEATEC awards. NEDO, the New Energy and Industrial Technology Development Organization, said on October 6 that Direava’s surgery-specific generative AI, a Surgical VLM, received the Innovation Award at CEATEC AWARD 2026. The model was developed under GENIAC, the generative AI programme NEDO runs with the Ministry of Economy, Trade and Industry.

The model analyses endoscopic surgery video and puts into words what is in the surgical field, what action is being performed and the surgical knowledge behind it. NEDO describes the shift as moving AI from the surgeon’s “eyes,” recognising organs, vessels and nerves, toward something closer to a “brain” that interprets the situation. Direava says it is the first foundation model of this kind in Japan.

Surgeons themselves wrote the question-and-answer training data, organised into four categories: object, action, reason and plan for the next step. The team then fully fine-tuned a large Japanese multimodal model on it. Because no evaluation standard existed, Direava also created its own Surgical-VLM Bench, in which specialist surgeons grade answers on a five-point scale. In a real-time trial in a university hospital operating room, the model scored 84.7 percent for anatomical accuracy and 82.9 percent for clinical usefulness.

What it means in practice

Direava’s Surgical AI Model Wins CEATEC Award 2026 Innovation Prize — contextual photo

The case for the technology is demographic. NEDO cites forecasts that the number of gastrointestinal surgeons will roughly halve over the next 20 years, and that by 2040 Japan will be about 5,000 doctors short for cancer surgery. Direava plans to start with surgical education for young surgeons, medical students and regional core hospitals. Only later, after the necessary regulatory procedures, does it intend to look at presenting information during operations, and eventually at partial automation.

Staff shortages are a recurring driver of AI adoption in Japan, from companies turning to AI robots to TEPCO’s trial of generative AI for nuclear paperwork. Medicine has its own track record too, as in hospital networks sharing playbooks for imaging triage.

What to watch: the demonstration at NEDO’s booth during CEATEC 2026 at Makuhari Messe from October 13 to 16, and Direava’s next GENIAC project, a model that predicts how an operation will unfold in order to improve surgical safety.

← Back to AI4Japan