Characterisation, identification, clustering, and classification of disease.

Area of Science:

• Biomedical Informatics
• Epidemiology
• Genetics

Background:

• Quantifying multimorbidity distribution and determinants is crucial for novel data-driven disease classifications.
• Existing studies utilize molecular information, disease incidence age, and disease trajectories for disease clustering.
• The potential of easily measured risk factors for disease characterization remains underexplored.

Purpose of the Study:

• To investigate if easily measured risk factors (height, BMI) can characterize diseases in UK Biobank data.
• To apply rigorous statistical methods for clinically relevant disease comparisons and clusters.
• To explore the utility of risk factors in understanding disease relationships and multimorbidity patterns.

Main Methods:

• Analysis of over 400 common diseases using clinical and epidemiological criteria.
• Application of conventional proportional hazards models to assess associations with 12 established risk factors.
• Utilizing UK Biobank data for risk factor analysis and disease clustering.

Main Results:

• Sex-dependent associations between disease risk and BMI were observed for several diseases.
• A significant proportion of diseases in both sexes were identifiable by their specific risk factors.
• Diseases with similar risk factor profiles tended to cluster together, including 10 diseases classified as 'Symptoms, signs, and abnormal clinical and laboratory findings, not elsewhere classified'.

Conclusions:

• Easily measured risk factors can effectively characterize and cluster diseases, providing a new perspective on disease interactions.
• Identified disease clusters often share known or suggest potential unconfirmed pathogenic mechanisms.
• This approach offers novel insights into the interplay between biological pathways, risk factors, and multimorbidity patterns.