Development of Surgery-Specific Generative AI Bringing a 'Brain' to the Operating Room — Successful Demonstration in Surgery, Full-scale Launch Toward Commercialization
Direava Inc. has developed Japan's first surgery-specific generative AI, 'Surgical VLM', capable of understanding real-time surgical contexts and interacting with doctors. Proven successful at Keio University Hospital, it aims for commercialization in 2026 to support medical education and improve quality.
📋 Article Processing Timeline
- 📰 Published: March 30, 2026 at 21:29
- 🔍 Collected: March 30, 2026 at 22:56 (1h 26m after Published)
- 🤖 AI Analyzed: April 24, 2026 at 06:55 (583h 58m after Collected)
Direava Inc. has undertaken the development of a 'Visual-Language Integrated AI Foundation Model for Surgical Support' as part of the GENIAC (Generative AI Accelerator Challenge) project. Implemented by NEDO and the Ministry of Economy, Trade and Industry, GENIAC aims to strengthen generative AI development capabilities. Direava has now successfully developed Japan's first surgery-specific generative AI (Surgical VLM) that understands surgical situations in real-time and communicates interactively.
Conventional generative AI used in the surgical field has been limited to assisting in the recognition of patient organs and affected areas through image recognition (functioning as 'eyes'). The newly developed surgery-specific generative AI is a groundbreaking system that understands 'what to do next' based on the surgical situation, acting like a 'doctor's brain,' and engages in advanced dialogue with surgeons.
On February 20, 2026, an empirical test was conducted during an actual gastric cancer surgery at Keio University Hospital, successfully confirming its usefulness by meeting the required standards of medical education (anatomical accuracy, clinical usefulness, and fluency of text).
Recently, training young professionals who inherit advanced medical techniques has become a critical challenge in the medical field. Moving forward, Direava will begin full-scale preparations for commercialization (service launch) in 2026 and will proceed with demonstrations in surgeries other than gastric cancer. This will strongly support the training of surgeons and advanced surgical education, contributing to the improvement of future medical quality and ensuring safety.
## 1. Background
Since FY2024, NEDO has been promoting the GENIAC project alongside the Ministry of Economy, Trade and Industry to accelerate R&D and support for generative AI. GENIAC provides support for the development of domain-specific foundation models derived from general-purpose models, with 53 companies developing generative AI foundation models as of the end of March 2026.
Direava, which achieved this milestone, is one of these companies and a medical startup originating from Keio University School of Medicine. It has been developing medical AI and related systems, having obtained manufacturing and marketing approval from the Minister of Health, Labour and Welfare on December 9, 2025, for its AI-equipped programmed medical device, 'Surgical Video Recognition Program Kinosura' *1.
Selected for GENIAC in July 2025, the company developed Japan's first*2 surgery-specific generative AI 'Surgical VLM'. By training a massive dataset of intraoperative images and natural language captions based on its proprietary foundation model, it implemented unprecedented functions. The results were demonstrated in actual surgical settings. In evaluating the developed AI, with the cooperation of Keio University Hospital, criteria were established not only for accurate assessment of surgical situations but also for factors deemed crucial for training surgeons.
## 2. Current Achievements
### (1) Development of Generative AI with a 'Brain' that Understands Surgical Context
Until now, the application of generative AI in surgery has progressed in 'anatomical recognition (= visual function)' to identify patient organs and blood vessels using image recognition. However, a major challenge has been the lack of implementation of 'advanced situational understanding (= brain function).' This capability is necessary to infer and converse about 'what to do next' and 'what to be careful of' based on recognized information while understanding complex surgical procedures and the state of the affected area.
Against this backdrop, Direava continues to innovate for medical education sites.
Conventional generative AI used in the surgical field has been limited to assisting in the recognition of patient organs and affected areas through image recognition (functioning as 'eyes'). The newly developed surgery-specific generative AI is a groundbreaking system that understands 'what to do next' based on the surgical situation, acting like a 'doctor's brain,' and engages in advanced dialogue with surgeons.
On February 20, 2026, an empirical test was conducted during an actual gastric cancer surgery at Keio University Hospital, successfully confirming its usefulness by meeting the required standards of medical education (anatomical accuracy, clinical usefulness, and fluency of text).
Recently, training young professionals who inherit advanced medical techniques has become a critical challenge in the medical field. Moving forward, Direava will begin full-scale preparations for commercialization (service launch) in 2026 and will proceed with demonstrations in surgeries other than gastric cancer. This will strongly support the training of surgeons and advanced surgical education, contributing to the improvement of future medical quality and ensuring safety.
## 1. Background
Since FY2024, NEDO has been promoting the GENIAC project alongside the Ministry of Economy, Trade and Industry to accelerate R&D and support for generative AI. GENIAC provides support for the development of domain-specific foundation models derived from general-purpose models, with 53 companies developing generative AI foundation models as of the end of March 2026.
Direava, which achieved this milestone, is one of these companies and a medical startup originating from Keio University School of Medicine. It has been developing medical AI and related systems, having obtained manufacturing and marketing approval from the Minister of Health, Labour and Welfare on December 9, 2025, for its AI-equipped programmed medical device, 'Surgical Video Recognition Program Kinosura' *1.
Selected for GENIAC in July 2025, the company developed Japan's first*2 surgery-specific generative AI 'Surgical VLM'. By training a massive dataset of intraoperative images and natural language captions based on its proprietary foundation model, it implemented unprecedented functions. The results were demonstrated in actual surgical settings. In evaluating the developed AI, with the cooperation of Keio University Hospital, criteria were established not only for accurate assessment of surgical situations but also for factors deemed crucial for training surgeons.
## 2. Current Achievements
### (1) Development of Generative AI with a 'Brain' that Understands Surgical Context
Until now, the application of generative AI in surgery has progressed in 'anatomical recognition (= visual function)' to identify patient organs and blood vessels using image recognition. However, a major challenge has been the lack of implementation of 'advanced situational understanding (= brain function).' This capability is necessary to infer and converse about 'what to do next' and 'what to be careful of' based on recognized information while understanding complex surgical procedures and the state of the affected area.
Against this backdrop, Direava continues to innovate for medical education sites.