Ayesa-Ibermática drives (r)evolution in healthcare data systems through ethical machine learning

As part of its efforts to develop innovative digital and AI-based tools, technologies and solutions, the company is developing a platform to improve how medical information is processed, exchanged and managed.

The solution is based on a high level of data privacy and ethics.

Ayesa-Ibermática, a global provider of technology and engineering services, and its i3B Institute for Innovation, is currently working with BioBizkaia and its other partners on the AISym4Med innovation project. The aim is to develop a platform based on machine learning able to provide healthcare data engineers, practitioners and researchers with access to a synthetic dataset system that is scalable, much more secure and reliable. This information is expected to prove highly useful for experimenting with and creating AI models for the healthcare sector.

The initiative is particularly important in that it will allow disease risk factors to be defined, transmission simulated and controlled experiments carried out using data that is simple to gather and reliable. In terms of accessibility and potential for reuse, it also allows the scientific community to fully comply with ethical and legal requirements, particularly when working with medical data and personal information.

The solution will guarantee information security and privacy by combining new data anonymisation techniques, attribute-based privacy measures and reliable tracking systems. It will also use artificial intelligence to apply a controlled data synthesis methodology for experimentation and modelling purposes.

Data quality

Another challenge is that clinical data can be incomplete, lack quality and fail to follow a standard format. Additionally, having data spread across different hospitals, clinics and government databases makes it difficult to use for research purposes.

As a result, Ayesa-Ibermática will also implement a series of data quality control measures. These will ensure unbiased data and respect for ethical requirements, and involve contextual searches as well as a human-centred design for validation purposes. The ultimate aim will be to ensure that the synthetic data generated is a representative sample. In this regard, one of the modules will be responsible for exploring and developing synthetic data creation techniques even further, dynamically supplying data on demand and for specific uses.

The platform will also use various technologies to reproduce unidentifiable data. This will allow for the indirect evaluation of a greater number of databases, while meeting privacy, security and GDPR requirements.

As R&D Director Itziar Cuenca explains, ‘the proposed framework will support the development of innovative and unbiased AI-based tools, technologies and digital solutions, which will bring a range of benefits for researchers, patients and healthcare providers, whilst maintaining a high level of data privacy and ethics’.

She also notes how AISym4Med will help to create ‘more robust machine learning algorithms based on the most effective computation configuration, and a meta-engine will provide information on the quality of the generalised model’.

The platform, which will ultimately improve the way data is processed, exchanged and managed, particularly for medical research, will be validated through local, national and international use cases for data engineers and ML developers.

The project is funded by the EU under the Horizon Europe programme and aligned with the initiatives of the European Commission, which actively supports the creation of a European Health Data Space. The purpose of this is to make it easier to exchange data and support research into new preventive measures, treatments, medicines and medical devices, whilst ensuring individuals have full control over their personal data.