Mediwork AI
Mediwork AI provides an integrated operational system for medical AI data construction, model development, evaluation, and clinical validation, powered by a network of 1,300+ physicians across 25 clinical departments. The company quantifies labeling quality using agreement metrics like Dice, IoU, and Krippendorff's α, and delivers traceable quality documentation to support regulatory approval submissions.
- Artificial Intelligence
- Healthcare Technology
- Software Only
Funding
Founders
Founder details are not available yet.
Product
Problem
Building high-quality training and evaluation datasets for medical AI development requires specialized medical knowledge, yet structured collaboration between healthcare institutions and AI companies is lacking, and objective methods to verify data quality are limited. Verification of labeling workers' qualifications and the review process are not systematically managed, making it difficult to secure evidence documentation for regulatory approval submissions.
Solution
Mediwork AI provides an integrated operational system covering the entire process from medical data construction to model development, evaluation, and clinical validation, based on a network of 1,300+ medical professionals across 25 clinical departments. Physicians who meet task-specific qualification requirements perform labeling and review, with quality quantified using agreement metrics such as Dice, IoU, and Krippendorff's α. The company delivers datasets along with quality metrics, review histories, labeling guidelines, and quality management documents to support both medical AI development and regulatory approval submissions. AI pre-labeling is used to improve efficiency and consistency of repetitive work, and the company also performs medical LLM response evaluation and RLHF/DPO preference data construction.
Target Audience
Primary customers are medical AI companies, medical device and pharmaceutical firms, domestic and international big tech and AI companies, and university hospitals, research institutions, and public organizations that need medical data construction and AI development and validation.
Features
- Network of 1,300+ physicians across 25 clinical departments with rapid task-appropriate matching based on specialty qualifications
- Medical data construction: expert physician labeling and review of imaging, text, and signal-based AI training and evaluation datasets
- Medical AI development: classification, segmentation, detection, and prediction models, plus RAG and medical imaging synthetic data generation
- Model evaluation and validation: medical LLM response evaluation, RLHF/DPO preference data, performance evaluation and clinical validation support
- Quality management system: qualification, judgment, review, and change history tracking, with Dice, IoU, and Krippendorff's α agreement metrics
- Traceable QA: license verification of participants, project-specific qualification checks, and AI pre-labeling for efficient workflows
- Research planning and consultation: IRB/DRB support, study design, and collaborative projects with medical institutions and specialists
- Web-based labeling platform with pathology workspace and live annotation demos