As a significant global public health problem, chronic kidney disease (CKD) is a progressive disease with no cure and a long disease course. Patients with CKD often co-occur with other diseases and exhibit notable clinical heterogeneity due to a wide variety of causes and varying disease trajectories over time.
Leveraging advancements in machine learning methodologies and the abundance of healthcare data, our research can separate CKD patients into groups based on their disease trajectories generated from diagnostic records. Our objective is to reconcile the exhibited heterogeneity of CKD patients across their disease trajectories, highlight demographic and clinical characteristics (e.g., mortality, age of onset, comorbid conditions at enrollment) among CKD patient subgroups, and identify the potential prognostic factors associated with favourable or adverse outcomes.
Past cluster analyses of CKD have certain limitations, and many studies have proven machine learning to be powerful and flexible for cluster analysis across various diseases (e.g., systemic lupus erythematosus, type 2 diabetes mellitus, Parkinson’s disease), helping capture the complexity of disease progression and reconcile disease heterogeneity. To date, there have been no reports of machine-learning approaches to establish clusters of CKD patients using disease trajectories, and therefore, our research holds substantial innovation and potential.
Given the magnitude of data involved and the intricacies of analysis, we anticipate a project duration of one to three years. We believe that the insights garnered from our study have the potential to provide scientific evidence supporting early detection, personalised diagnosis, and tailored management strategies, thereby improving patient outcomes and alleviating the public health burden associated with CKD.
