Humans’ behaviour is at the centre of the capability of Personal Health Data Science to impact evidence-based outcomes with self-reported data as a fundamental asset for that evidence. However, its use within machine learning processes is challenging, because self-reported data is noisy, volatile, and diverse while actual machine learning algorithms need clean, classified, and homogenous data, preferably in a silo, and within a single dimension approach.
The challenge for machine learning is then to be able to use self-reported data to allow a better understanding of individual health in a process from population to the individual. To use this data, we will focus on developing machine self-learning algorithms that overcome the human need to classify data and to build this we are going to start from knowledge graphs produced using machine self-learning capable to describe diseases, such as diabetes, obesity, dementia, or aged related macular degeneration (AMD) and aim for the development of a machine self-learning approach to support personalized decision making.
This would allow the development of policies and frameworks fitted to change behavior which is a central element to reduce the burden of these diseases on the health care systems. With this project we will develop the machine self-learning algorithms to create personal evidence on individual multiple long-term conditions (MLTC) also known as multimorbidity’s.
This project will be carried out in cooperation with machine learning and AI scientists at SANO and its partners and by developing external collaborations with Krakow health providers and with UK and Portuguese Universities.