Project description
Machine learning identifies risk factors for diabetes
Diabetes affects millions of individuals worldwide, and the inability to effectively manage blood glucose levels leads to various comorbidities such as obesity and cardiovascular disease. Therefore, it is important to understand patient characteristics and develop health interventions that stimulate positive health behaviour changes. To achieve this, the EU-funded CASCARA project will employ a multi-parametric approach that includes machine learning to identify specific characteristics capable of predicting how patients respond to diabetes diagnosis. Researchers will use observational data to determine the impact of gender and socioeconomic and demographic status on the adoption of a healthy lifestyle that involves a well-balanced diet and physical activity.
Objective
Diabetes causes a large and unevenly distributed health and economic burden within the population living with diabetes. Improved health behaviours have the potential to avert a large share of morbidity and mortality attributable to diabetes. However, adherence to recommended self-management remains challenging for many patients. This may (at least partly) explain the large overall disease burden in people with diabetes, as well as how that burden is distributed among patients. A better understanding of the patient and community level characteristics that affect behaviour change can inform more personalised, more effective health interventions that stimulate positive health behaviour changes, in turn reducing the overall burden associated with diabetes.
CASCARA aims to provide novel and much needed evidence on characteristics predictive of (1) health behaviour change subsequent to a diabetes diagnosis and (2) of the resulting changes in diabetes complication risk factors. To achieve this, I will use causal econometric and epidemiologic methods as well as machine learning (ML) and causal mediation analysis. The commonly recommended behaviour changes I focus on comprise: improving diet, increasing physical activity, reducing smoking and alcohol consumption. In particular, CASCARA will address the following research objectives using longitudinal observational data from continental Europe, the UK and the US:
1. Investigate the effect of a diabetes diagnosis on health behaviours and potential heterogeneities across gender and socioeconomic status
2. Use of ML to identify potentially unanticipated socioeconomic, demographic and clinical characteristics affecting health behaviour change, for a more detailed understanding of its potential drivers
3. Use causal mediation analysis to identify the impact of different health behaviour changes on risk factors for diabetes complications (body mass index, hypertension status and blood glucose levels) post-diabetes diagnosis.
Fields of science
- social sciencessociologydemographymortality
- medical and health sciencesclinical medicineendocrinologydiabetes
- medical and health scienceshealth sciencesnutrition
- natural scienceschemical sciencesorganic chemistryalcohols
- natural sciencescomputer and information sciencesartificial intelligencemachine learning
Keywords
Programme(s)
Topic(s)
Funding Scheme
MSCA-IF - Marie Skłodowska-Curie Individual Fellowships (IF)Coordinator
4366 Esch Sur Alzette
Luxembourg
The organization defined itself as SME (small and medium-sized enterprise) at the time the Grant Agreement was signed.