Synthetic Data Services
Overview
The proposed service provides a privacy-focused solution that specializes in advanced anonymization of structured data and AI-driven synthetic data generation. Data anonymization transforms sensitive datasets into fully de-identified, regulation-compliant assets while preserving their statistical accuracy and analytical value. Synthetic data generation creates realistic datasets with artificial data that mirror real-world patterns without exposing any personal information or compromising privacy. This service helps to unlock the full potential of data while maintaining security and compliance standards, ideal for testing, validation, benchmarking, and generating data for AI training. For processing sensitive personal data in structured format and anonymization evaluation, the service uses tools such as ANJANA, pyCANON, CTGAN, SynthCity, ydata-synthetic, and TSHM, supporting GAN-based generators, VAE-based generators, diffusion models, and autoregressive/sequence models. For generating synthetic data for kinematic analysis, the service combines 3D rendering software (e.g., Blender) with generative neural network models to produce realistic synthetic images that conform to specified pose configurations. Synthetic datasets are generated by leveraging pose annotations from widely used benchmark datasets, including MS COCO and MPII. In addition to fully synthetic generation, synthetic data is also produced through image augmentation techniques applied to existing images, further increasing dataset diversity and supporting the development of more robust models.
Outcome
Anonymized data, data generation models, and generated exemplar synthetic datasets in pose estimation and health domains, including synthetic per-person aggregated behavioral markers and visit-level sequences for risk prediction, synthetic pain datasets with observable cues including facial expressions and multimodal physiological signals, synthetic breast cancer recurrence datasets, and synthetic human pose datasets generated from MPII and MS COCO keypoints.
Service Card
| Category | Infrastructure and data-related services |
| Domain | Data Privacy, Synthetic Data Generation |
| Vertical | Cross-domain |
| Target users | SMEs, Public Sector Organizations, Data Analysts, Data Engineers |
| Service type | Service |
| Outputs | Anonymized datasets, trained synthetic data generation models, and exemplar synthetic datasets in health and pose estimation domains. |
| Requirements | |
| Limitations | |
| Availability | Q3/2027 |
| Licensing | |
| Visual content | assets/service-visuals/synthetic-data-services.png |
| Visual caption |