Endotypes and Trajectories of Multi-morbidity for Stratification and Biomarker Analysis
Many patients seeking treatment have more than one condition or illness. Medical professionals have developed the term "comorbidity" for the presence of more than one disorder in an individual. These can include cardiovascular disease, cancer, inflammatory diseases, arthritis, mental health issues and dementia. Comorbidities affect one in five adults and two-thirds of the elderly. This high prevalence presents challenges for both the patient and healthcare providers who often are geared to consider just one disease at a time. When comorbidity has been studied effectively it has helped treatment strategies, for example in sepsis, asthma, and acute respiratory distress syndrome. Our aim is to understand comorbidity more fully, identify risk factors for comorbidity development, and to find biological molecules (easily measured in blood) that predict or define disease.
Using the computational power and the techniques that fall under an umbrella term of artificial intelligence (AI) we will analyse the vast data from UK Biobank on volunteers who have developed comorbidities, to understand how these accrue and how to predict their development more fully for patient benefit in the future.
Our plan, then, will be to use computational algorithms to sort UK Biobank volunteers into groups with different combinations of comorbidities, and progression in developing and accumulating comorbidities, based on their healthcare records. We will then identify differences in clinical, health, environment and socioeconomic characteristics between these groups of patients. Next, we can look at the complex array of molecules (such as proteins and lipids) in their blood to determine if blood-based biochemicals can be used to predict disease development or current status. We will also look for signatures in the DNA that associate with those UK Biobank volunteers who sadly suffer from comorbidities.
Our plan for assessment of UK Biobank comorbidity grouping and the relationship to presently acquired data on participants will span a 3-year period. We will share our findings from our AI-based work on clustering comorbidity patients in the form of scientific publications. Results from this project will enhance our effective understanding of different diseases in the same person, with huge potential to improve patient management and care through better stratification and the development of tools that improve predictions for patients.