Medical data plays a key role in modern healthcare. However, its accessibility is limited by regulatory and technical barriers. Data is often stored in small volumes, scattered across multiple healthcare organizations, in unstructured form, or in proprietary data formats of various Medical Information Systems.
These issues, along with the lack of interoperability between systems and difficulties in the timely collection of routine clinical practice data (RWD), create obstacles for all participants in the medical process:
Pharmaceutical companies and researchers cannot effectively evaluate the therapeutic effect of drugs and conduct marketing research.
Developers of AI-based CDSS face difficulties in training and testing models on high-quality labeled data.
Doctors do not receive timely and interpretable feedback on the compliance of prescriptions with clinical guidelines.
The RWD Marketplace project addresses these issues by creating a free market for anonymized medical data, enabling more accurate marketing and pre-clinical research, accelerating the development and testing of CDSS, and supporting the development of data science and machine learning technologies.
How it works
We work with private healthcare organizations based on one of the following principles:
For clinics that already have a Medical Information System
We offer an integration bus for medical data exchange that not only collects data but also performs several key functions:
Complete anonymization of data in accordance with local legislation.
Automatic structuring of medical data into a structured electronic medical document without a predefined template.
Creating RWD datasets for further analysis.
For clinics that use our Medical Information System…
…and are interested in RWD commercialization, no additional action is required — the clinical repository will already contain high-quality structured medical data
For data consumers
We provide the ability to configure the following parameters:
Required fields (e.g., gender, age, ICD diagnosis, drug prescription according to International Nonproprietary Name / Trade Name, etc).
Filtering by these fields.
Download size and geographical restrictions.
Ability to customize the dataset structure, including complex fields and custom entities.
Uniqueness of the solution
We can work with any type of data, including unstructured and proprietary formats from various Medical Information Systems. We combine a non-ML approach for initial data labeling with an ML approach for subsequent extraction and labeling of anonymized medical data. This allows us to significantly automate the process and connect the “long tail” — small medical organizations that previously were unable to participate in such projects due to insufficient data volumes and high data collection costs.