COTADS

COTADS COdesigning Trustworthy Autonomous Diabetes Systems

Health & Wellbeing

Summary

COTADS will explore how to increase trust of AI used for diabetes management inside and outside clinical settings during life transitions. AI is expected to provide a crucial role in the management of chronic conditions, yet technology-driven solutions are unlikely to be adopted. AI design must consider the complex medical, lifestyle and socio-technical needs at times of uncertainty and life transitions. COTADS will bring together people with diabetes, clinicians, and data scientists in a novel co-design process for diabetes risk stratification. Using co-design, provenance, and explainable AI, we aim to ensure solutions are understandable, transparent, trustworthy, and beneficial.

Objectives

COTADS brings together participatory co-design, provenance, and explainable AI in the design of risk stratification models for diabetes management inside and outside of clinical settings at times of life transition.

We will build on our novel research in co-design for AI by understanding the potential for participatory processes to aid care both inside and outside clinical settings, as both are key to the successful management of this idiosyncratic condition. Knowledge extracted from co-design will be augmented with AI pipeline provenance, increasing accountability/traceability of requirements, whilst finally, explainable machine learning models will be used to interpret risk stratification. These techniques aim to deliver novel human-AI interactions for decision making and enhance accountability, transparency, and trust of digital interventions supporting diabetes management.

The video below provides an overview of the COTADS project objectives