Trusted AI for Diabetes personalised self-management systems

About the Project

Diabetes mellitus is a common cause of morbidity and mortality across the globe. Type 2 diabetes continues to rise in prevalence annually; from 1980 to 2004, Type 2 quadrupled in both prevalence and incidence [1]. This will be exacerbated further by the increase in obesity. It is estimated that diabetes will be the seventh leading cause of death worldwide by 2030 [2].

Type 2 diabetes has a genetic influence, but is also strongly linked with food consumption and a sedentary lifestyle. The costs of managing diabetic complications currently outweigh the costs of anti-diabetes drugs by a factor of 3-4 [3]. Research has shown that improved self-management could be a powerful tool in reduction of morbidity and mortality from type 2 diabetes, whether from prevention or improved glycaemic control. Research that can inform policy and guide management has the potential for large cost savings and to make a substantial impact on public health.

Big data collected from Diabetic individuals over the years can be used to provide personalised treatment plan to such individuals and informed decision support to healthcare providers and policy makers.  Type of such data includes:

• Electronic health records
• Smart glucose meters
• Insulin pumps and automated insulin delivery system
• Patient-held data e.g. diabetes apps
• Digital images from retinal screening

The increasing use of real world data will ensure that the evidence applies to the patient group in question. Factors peculiar to a particular ethnic group, locality or institution will be rapidly detected. This will enable the benefits of research to apply to the entire patient population and will particularly benefit ethnic minorities whose relevant genetic characteristics or other factors may differ from the general population.

Artificial intelligence (AI) along with the increased availability of health data hold great potential to pave the way for personalised prevention and enable progress towards risk prediction and early detection of chronic non-communicable diseases

Development of AI tools to explore such data and identify insight and knowledge to personalize treatment plan and prediction of complications to support clinicians while treating their patients can result in better health outcomes and reduction in cost.  AI tools can also be developed to be part of the self-management apps that the patients use to support in managing their lifestyle and diet decisions.  AI tools can also be used by policy maker in planning for the management of resources for such diseases.

Big Data projects require an appreciation of both the ethico-legal milieu and the socio-political landscape. Addressing such aspects is vital to develop trust in using AI tools in the healthcare domain.

The developed AI tools will be based on the FAIR [4] data principles and the good practices for GDPR-compliant personal data protection. The solution will also be based on international standards and best practices used in the development of AI solutions.

Funding Notes

there is no funding for this project