An ecosystem for collaboration hospitals and pharmaceutical companies on AI projects

The core elements we provide:

  • Health data analytics (data mining, data aggregation, data modelling, etc.)
  • Artificial intelligence (AI) solutions for treatment and care (computer vision, investigation of surgical workflows, knowledge representation, learning from data).
  • Annotation/labelling for video (e.g. surgical procedures) and static data (e.g. histopathology, x-rays, etc.), ensuring high quality of annotations and data preparation for machine learning (ML).
  • Data anonymization, data reuse, and data security.
  • Situation awareness, reasoning and decision support techniques.
  • Complex analysis of AI and ML solutions for healthcare, including deep learning algorithms, recommendation systems and end-to-end decision-making (DM) systems.

Our research group proceeds a large amount of multi-modal data from electronic health records, cameras, and medical devices to develop AI-based systems to assist clinicians and staff in their medical routines.

To ensure privacy and security while training ML models, we utilize rigorous data preparation alongside with series of data protection techniques in the federated learning pipeline as data aggregation, perturbation, cleaning.

Early stage clinical trials: Current Stage

Individual Case Safety Reports (ICSR)

Extracting ICSR specific information is one of the biggest challenges facing the industry
  • ICSR involves understanding the complete text and entering into the case, the details of the adverse event, drugs, patient data, his/her history and so on


  • data sharing
  • store user data
  • volume growth
  • privacy and security of sensitive data

The goal:

to develop a best-in-class platform that connects clinical and non-clinical data to accelerate drug discovery and treatment advancements for patients


  • Customer engagement
  • Privacy and Security
  • Accuracy andEfficiency

Tools and technologies:

  • Distributed ledger
  • Federated learning
  • Machine learning and AI

State-of-the Art

Tools and enterprise solutions around Federated Learning and other secure computation techniques across different verticals

TensorFlow Federated

an open source framework by Google for experimenting with machine learning and other computations on decentralized data


an open source library that is built on top of PyTorch for encrypted, privacy preserving deep learning


(Federated AI Technology Enabler), an open-source project initiated by Webank’s AI group to provide a secure computing framework to support the Federated AI ecosystem


is one of the pharma company taking a lead in implementing AI-based solutions in drug discovery

Federated Learning Startups

Drug design startups

What we propose

Federated biomedical research and healthcare AI ecosystem

Artificial Intelligence-powered platform for drug discovery and collaboration pharmaceutical companies, hospitals andindividual users Source

Three core components

DLT + Federated Learning + AI

Orchestrated Customer Engagement

  • Our model enables more effective customer engagement
  • It can engage different stakeholders include payers, providers and
    influencers, which range from national healthcare organizations to
    diagnosticians, pathologists and health informatics experts.
  • It also includes bodies like clinical commissioning groups, specialized
    commissioning and specialist trusts; as well as Cancer Alliances and
    primary care hospitals.

Support full-stack privacy and security

  • We use federated learning and homomorphic encryption to keep sensitive
    data private.
  • Ecosystem based on a distributed edge computing infrastructure.
  • Train an algorithm across multiple decentralized edge devices or servers
    holding local data samples, without exchanging their data samples.
  • These enable to reduce cloud infrastructure overheads and optimize all
    available resources for efficient and economical scaling

Acceleration meaningful healthcare outcomes

  • drug-discovery algorithms on each-other’s data
  • algorithms might be trained on a pharma company’s data but data
    itself wouldn’t be shared with collaborators instead, the models from
    each dataset are shared and averaged together in the central model
  • model development, training, and evaluation with no direct access to
    or labelling of raw data, with communication cost as a limiting factor

Why it works

Speed of deployment

  • immediate access to data
  • ability to reuse data

Operational flexibility

works with multiple machine learning frameworks and data pipelines

Risk mitigation

sensitive data is non-identified throughout the research process

Deploy or pilot new capabilities

  • faster and more efficient drug research, more robust evidence to support treatment value, and greater access to personalized medicines
  • apply machine learning to identify and screen potential drug candidates

Bring in expertise in therapy area

  • data encourages sharing between institutions by addressing privacy and competitive concerns

Sunderland Software Centre
Tavistock Pl, Sunderland SR1 1PB,
United Kingdom

6 Strokacha str. Kyiv, Ukraine 03148
+380 44 596 41 51