• 151

    Projects submitted

  • 109

    Published projects

  • 47

    Comments left

  • 66

    Number of votes

  • 52

    days until voting ends

← Back to list

Kazakhstan's first regulated platform for de-identified medical data

  • Project manager on customer side

    Togjan Qojaly

    MedHub 'Health AI Quality'

    Founder and CEO

  • Category

  • Nomination

  • Goals

    Medical data in Kazakhstan was fragmented across disconnected hospital systems, limiting its controlled use for research and AI development.

    The project pursues four key objectives:

    1. Establish a platform for exchanging de-identified, quality-checked medical data products while ensuring that raw clinical data remains with its original holder.

    2. Register the platform under Kazakhstan’s regulatory framework for data exchange platforms.

    3. Enable data holders to retain control over their data through traceable transactions, while providing data users with verified data products and a clear legal basis for access.

    4. Establish the data infrastructure required for AI testing and clinical validation of medical AI solutions.

    As of July 2026, the platform was registered, with one data owner connected and 30 data products published, including 29 free products and one paid product.

  • Timeline

    june, 2026 — july, 2026
  • Project completion year

    2026

  • Project scale

    600 human-hours
  • Results

    1. Regulatory milestone: on 20 July 2026 the platform was entered in Kazakhstan's national registry of data exchange platforms as Platform No. 1 (No. KZ23TZY00000001), one of the first regulated medical data markets in Central Asia.

    2. Platform launched: the full data product lifecycle works in production, from de-identification at the source and an isolated quarantine zone to a 15-step quality check, data passports and an immutable transaction log.

    3. Catalogue: 1 data owner connected and 30 data products published (1 paid, 29 free).

    4. Before / after: previously a researcher or AI developer had no lawful channel to obtain de-identified clinical data, and transfers were informal and unverified. Now a verified dataset with a clear legal basis can be selected from a public catalogue.

    5. Efficiency: the platform was delivered in about 600 person-hours on a stack of mostly free and open-source tools.

    The first commercial transactions are the next milestone.

  • Project uniqueness

    The project pioneers a regulated medical data market in Central Asia, built as Kazakhstan’s new legal framework came into force in July 2026, recognizing datasets as objects of civil rights.

    Raw and personal data never leaves the healthcare provider. De-identified data enters an isolated quarantine zone with no export, undergoes 15-step input control and quality checks, and receives a data passport documenting its origin and quality. K-anonymity thresholds apply (k≥5; k≥11 for highly sensitive data), while every transaction is recorded in an immutable audit log.

    The platform creates a model serving all stakeholders: patients retain independent oversight, clinics turn underused data into auditable assets, researchers and AI developers access lawful, verified data products, and the state gains a transparent alternative to informal data transfers.

  • Software used

    Federated data storage; de-identification and anonymization; isolated quarantine environment; Data Warehouse / Data Lake; data catalogue with data passports; immutable transaction log; FHIR and HL7 interoperability standards; ISO 27001 / 27701 security and privacy practices; hosting on servers in Kazakhstan.

    Technology stack: TypeScript 5, Node.js 22, NestJS 11, Prisma 7, PostgreSQL 16, React 19, Vite 7, Tailwind CSS, Radix UI, TanStack Query, Keycloak 26, NCALayer, NCANode, ClamAV, Redis, Python 3, Ubuntu 24.04 LTS, nginx, Let’s Encrypt TLS, Postfix, Nodemailer, Gitea, Gitea Actions and Claude (Anthropic).

  • Solution from the Global CIO catalog

    The project does not use solutions from the Global CIO catalog

  • Project implementation complexity

    We built a market for something the law had only just recognised. When datasets became objects of civil rights in July 2026, the team designed the governance, contractual and technological foundations of a regulated medical data exchange while the rules were still taking shape. The platform became the first to be entered in Kazakhstan’s state registry.

    The harder barrier was trust. Medical data is among the most sensitive assets a healthcare provider holds, yet Kazakhstan lacked an established regulated marketplace for its controlled exchange. We addressed this challenge with an architecture designed to keep raw data within the clinic: de-identification at the source, an isolated quarantine zone with no export, a 15-step quality control process, k-anonymity thresholds and an immutable audit trail.

    Every data product had to meet requirements spanning law, security, medicine and commerce. Each became a verified asset with its own data passport, pricing and owner-defined access rules.

  • Description

    Context. Kazakhstan has strong clinics, researchers and a growing number of medical AI startups, but no infrastructure connecting them. Medical records sit in separate information systems. Direct "hand-to-hand" transfers carried high risk: no one verified de-identification, data escaped the owner's control once shared, no audit trail was kept, and no common standard existed.

    Solution. The team built a platform for the exchange of digital medical data products and registered it as Platform No. 1 in Kazakhstan's national registry of data exchange platforms (No. KZ23TZY00000001).

    How it works. Each data product moves through a controlled lifecycle:

    - de-identification at the source, before upload;

    - quarantine in an isolated zone from which nothing can be exported;

    - a 15-step input control and quality assessment;

    - a data passport with origin, quality and mandatory metadata;

    - publication in a public catalogue with published pricing;

    - access strictly on the data owner's terms;

    - an immutable, auditable transaction log.

    Architecture. Storage is federated: data stays in the clinic and only data products are exchanged. The platform is hosted on servers in Kazakhstan and follows FHIR and HL7, with ISO 27001 / 27701 practices and a DPIA audit. Security rests on k-anonymity thresholds and a ban on circulating raw or unprocessed data.

    Ecosystem. The platform is the core of a wider Health AI infrastructure: a data labelling office with physician annotation and quality control, an AI Sandbox for testing models, and a clinical validation centre. Resident startups already build on it, including an AI tool for cerebrovascular diagnostics on CT angiography and an AI assistant for lab test interpretation.

    Results. The platform was entered in the national registry on 20 July 2026. Since then, 1 data owner has been connected and 30 data products have been published in the catalogue (1 paid, 29 free), each with a data passport and a completed 15-step quality check. Before the platform, a researcher or AI developer had no lawful channel to obtain de-identified clinical data: transfers were informal and unverified. Now a verified dataset with a clear legal basis can be selected from a public catalogue. The first commercial transactions are the next milestone.

    Next steps. Within 12 months: 500,000+ medical records in the trusted data environment, 10 contracts and 5 data deals on the platform, and connection to AI validation in the sandbox.

  • Project geography

    Kazakhstan. The platform operates nationwide under the Digital Code of the Republic of Kazakhstan and is registered in the national registry of data exchange platforms as Platform No. 1. All data and infrastructure remain within the country: servers are hosted in Kazakhstan, raw data remains with the healthcare providers that own it, and platform operations are designed to comply with national requirements for personal data protection and digital signatures. The model is built to scale beyond a single city or hospital network. Healthcare providers across Kazakhstan can join as data owners, while researchers and AI developers can discover verified data products through a shared catalogue.Kazakhstan is the first step. Across Central Asia, healthcare systems face similar challenges, including fragmented medical data and the need for trusted, regulated exchange. A working model in Kazakhstan can provide a blueprint for neighbouring countries as their data governance frameworks evolve.

  • Additional presentations

Comment

Only authorized users can comment.
We suggest you or register.

  • Customer

    MedHub 'Health AI Quality'

We use cookies for analytical purposes and to deliver you the best experience with our website. Continuing to the site, you agree to the Cookie Policy.