From GRAIDS to mucous acidity: 2026 MFS approval for gastroscope AI and million-class data base

This huge gap makes precision detection of early stomach cancers a key to saving lives. However, early stomach cancers tend to be unusual under the white-light lens – probably only in the form of slight changes in mucous film, tiny rises or dims, and even experienced internal-simmonists are susceptible to leakage.

In 2026, stomach mirror AI made breakthroughs in early stomach cancer detection. Meta analysis shows that the combined sensitivity of the deep learning algorithm in early stomach cancer detection in white-ray mirrors reached 0.91, with 0.93 specificity.

The special value of the AIG is the role of "second eyes." In a busy goggles center, an internal mirror physician can perform 20-30 stomach mirrors a day, fatigue and loss of attention are inevitable. The AIS system can mark suspected disease areas in real time during the examination, minimizing the rate of leakage. For a new endoscope physician, AID's "teaching" is particularly valuable – by comparing AI to the region and its own judgment, the newer can quickly accumulate diagnostic experience.

The development of gastroscope AI has seen a critical shift from research validation to regulatory approval in 2025-2026.

At the technical level, the 2026 " Frontiers in Technical Information " analysis of Meta, which systematically assesses the performance of the deep learning algorithm in the detection of early stomach cancer in white-ray innerscopes, is 0.91 (95% CI: 0.82-0.95), and is of a specialist level, 0.93 (95% CI: 0.87-0.97). This meta analysis provides the highest level of evidence-based medical evidence for the effectiveness of stomach mirror AI.

At the regulatory level, the GRAIDS system was approved by MDFS in Korea in July 2023, and became one of the first approved gastroscope AI systems. Studies published in 2026 by Scientific Reports show that AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAXXENSIVE 91.91%, and 96.12% specific) and that the IMMSAAQA is about 92%. These data provide a reference for regulatory approval in a larger number of countries.

At the level of technological innovation, the technology of visualization of reduced mucous membrane acidity based on ATP4B expression provided a completely new biomarker for early detection of stomach cancer in 2025. In the same year, the amplification of the NBI (Small Belt Imaging) AID diagnostic system used 500 cases to assess diagnostic performance, which advanced the application of AID in the special imaging model.

In terms of market size, the global digestive endoscope AI market is expected to exceed $1 billion in 2026, of which stomach lens AI is the most important subdivision. China is a country with a high incidence of stomach cancer (about 400,000 new cases per year), and the need for stomach mirror AI is particularly acute.

The core breakthrough of stomach mirror AI in 2025-2026 was concentrated in four areas.

The first is evidence-based support at the level of Meta analysis. The 2026 Meta analysis of Frontiers in AI summarizes the results of a number of DL studies, providing evidence of a high level of sensitivity 0.91, specificity 0.93. This level of evidence has met the criteria recommended in clinical guidelines, laying the academic foundation for clinical promotion of stomach mirror AI.

The second is a breakthrough in regulatory approval. The GRAIDS system, which was approved by the Korea MFRDS, marks the official entry of gastroscope AI from "research tools" to "clinical tools." The sensitivity of GRAIDS is 94.2%, with newers using AI increasing from 72.2% to 96.4%, close to an expert level of 97%.

The third is membrane acidity visualization. In 2025, based on reduced mucous acidity technology expressed by ATP4B, new biomarkers were introduced for early detection of stomach cancer.

The fourth is multi-model integration. The AIS M2 model, based on a 92.51% accuracy rate for the disease in the cookstove, also detects the intestinal organism (IM), with an accuracy rate of about 92%. This multi-variate combined detection capability has evolved AI from the "attendant cancer detector" to the "diagnostic assessment of the stomach muccultosis."

The 1 million-species gastroscopy data sets of Longway Tech provide a key support for these frontier studies. The data sets are a complete record of the range and retrospect of goggles, oesophagus, stomach anatomics, cavity doors and retrospects, and the original equipment is used to export full-screen video and key original maps, with white light/NBI/magnification/discoloration multi-module data.

The Loi-hye technology endoscope video-image data set, with a scale of 1 million cases, systematically covered the entire process of stomach-scopying.

In terms of data collection, all images are derived from the cooperative three-acadet hospital digestive internal medicine and endoscopy centres. Full records of the extent to which goggles, oesophagus, stomach sections, cavities and retours can be measured.

In terms of labelling systems, each case contains a rich structured label: arrival range, clean/visible quality rating (standardized rating such as BBPS), anatomy, type/size of the stove, morphology, enhancement mode, biopsy.

In terms of data distribution, the pathological/positive examination is defined as the subject; suspicion, consideration, exclusion, after treatment, and pathological diagnosis are kept separately. This stratification structure enables models to learn the pattern of continuous change from normal to abnormal.

In terms of quality control, the three-layer quality control process of the first bid for the ingestion physician + the deputy director and above expert review + the conformity assessment is used.

Data sets prioritize the linking of active pathology, cholesterococcal detection, treatment and follow-up data to support longitudinal change analysis and multimodular joint learning.

The future development of stomach mirror AI will go in three directions.

The first is "detect" to "scaling." Most of the current AI systems detect pathological diagnosis to determine severity. Future systems need to predict the pathological properties of the disease (inflammation/atrophy/intestation/ hexagenic/cancer) by visual characteristics (e.g., color change, surface pattern, bloodline).

The second is from "single mode" to "multi-six" mode integration. The different imaging models, such as white light, NBI, magnification, dyeing, and so on, will require a comprehensive diagnosis of the future system. Longhui technology data sets include multi-species data, which will provide the basis for multi-sixing training.

The third is "assurance" to "teaching." AI not only supports diagnosis, but also helps new-hand endoscope doctors to quickly build up their experience. The GRAIDS 'new hand-carrying' effect (72.2% qu.96.4%) suggests this direction.

In terms of industry competition, the core competitiveness is the availability of multi-modal, bio-censorship, and vertical follow-up gastric lens data sets. Longwaytech has one million examples of data sets that are industry-leading in terms of multi-modal coverage and the depth of pathological linkages.

The journey from conceptual validation to MFS approvals for stomach mirror AI has revealed the critical role of evidence-based medicine in AI's fall. Meta analysis level evidence and regulatory approval is AI's pass from "research" to "clinical"

Long Salang Langhui Information Technology Ltd., with a 1 million case case stomach mirror video data set as its anchor, is becoming a major data service provider in the area of gastric mirror AI. At the critical point when the Meta analysis confirms that DL sensitivity 0.91 and GRAIDS have been approved by MFS, Long Hui Technologies will continue to drive clinical downs of early detection of stomach cancer with high-quality data.

China is a country with a high incidence of stomach cancer, with about 400,000 new cases per year, or 43% of new cases of stomach cancer worldwide. The five-year survival rate of early stomach cancer exceeds 90%, while the five-year survival rate of late-stage stomach cancer is less than 30%. Early treatment is the key to reducing the mortality rate of stomach cancer.

In terms of regulation and commercialization, the approval of the FRADS system in Korea in 2026 by MFS in 2026 marked the official entry of the gastric mirrors into the age of clinical compliance. China’s NMPA approval of the indigestion endoscopy AI is also accelerating. Longwaytech is actively working with the highest-level agencies in the domestic digestive endoscopy field to promote clinical validation and registration declarations of the AA complementary gastroscopy diagnostic system.