AB
AB
AiBoss

All Categories

project

Mingqi - A medical multimodal big data model launched by Shanghai Jiao Tong University for the accurate diagnosis of rare diseases.

Mingqi is the first large-scale medical multimodal model for the precise diagnosis of rare diseases launched by the LoCCS Lab of the School of Computer Science at Shanghai Jiao Tong University. Based on a dual-engine architecture of "large model capability matrix + expert routing collaboration," it integrates medical imaging...

What is Mingqi?

Mingqi is the first large-scale medical multimodal model for the precise diagnosis of rare diseases launched by the LoCCS Lab of the School of Computer Science at Shanghai Jiao Tong University. Driven by a dual-engine architecture of "large model capability matrix + expert routing collaboration," it integrates data from medical images, medical records, and laboratory indicators to provide doctors with high-precision and interpretable diagnostic support. In the diagnosis of rare gastrointestinal diseases such as Crohn's disease, its accuracy exceeds 92%, surpassing the level of senior specialists. Based on model distillation and quantization technologies, Mingqi significantly reduces computational power requirements, enabling low-cost, localized deployment and facilitating the downward flow of medical resources, bringing new possibilities for solving the problem of unbalanced medical resources in China.

The main functions of Mingqi

  • Precision diagnosisIt integrates multimodal data (images, medical records, and laboratory indicators) to achieve high-precision diagnosis of rare diseases (such as Crohn's disease), with an accuracy rate of over 92%.
  • ExplainabilityVisualizing the diagnostic process provides a basis for reasoning and comparisons with similar cases, thereby enhancing doctors' trust.
  • Multi-expert collaborationIt simulates the diagnostic approaches of multiple experts, integrates opinions from various parties, and improves the comprehensiveness and accuracy of diagnosis.
  • Localized deploymentThis reduces computing power requirements, enables low-cost and secure localized operation, and meets data privacy requirements.

Mingqi's technical principles

  • Dual-engine drive architecture:
    • Large Model Capability MatrixBased on large-scale pre-trained models, it learns features and patterns from massive amounts of medical data, providing a powerful foundation for the diagnosis of rare diseases.
    • Expert routing collaborationIt simulates the diagnostic thinking of multiple experts, integrates the experience and knowledge of different experts into the model, and realizes multi-expert collaborative diagnosis.
  • Multimodal deep learningThis system integrates multimodal data, including medical images (such as CT, MRI, and endoscopic images), medical record texts (such as medical history and symptom descriptions), and laboratory indicators (such as blood and biochemical indicators). Based on deep learning algorithms, it extracts key features from the multimodal data to achieve comprehensive diagnosis of rare diseases.
  • Transparent diagnostic cabin mechanismThe AI visualizes every step and reasoning logic in the diagnostic process, allowing doctors to clearly understand the basis of its diagnosis. It also provides a database of similar cases for comparison, helping doctors quickly locate suspected lesions and enhancing the interpretability of the diagnosis.
  • Model distillation and quantizationBased on model distillation technology, complex pre-trained models are compressed into smaller, more efficient models, reducing computational requirements. Based on quantization technology, the running efficiency of the model is further optimized, enabling efficient operation on low-cost hardware devices.

Mingqi's project address

Application scenarios of Mingqi

  • Precision diagnosis of rare diseasesIt assists doctors in quickly and accurately diagnosing rare diseases, shortening the diagnosis time.
  • Empowering primary healthcareTo provide low-cost, efficient diagnostic services to primary hospitals and alleviate the shortage of expert resources.
  • Teaching and Research SupportTo support medical teaching and research, enhance diagnostic capabilities, and promote technological development.
  • Telemedicine servicesBy combining telemedicine, remote diagnosis can be provided to patients in remote areas, improving access to healthcare.
  • Medical data managementIntegrating multimodal data facilitates storage, querying, and analysis, enabling the identification of disease risk factors.
Back