Tensor Factorization-Based Prediction With an Application to Estimate the Risk of Chronic Diseases

Abstract

Tensor factorization has emerged as a powerful method to address the challenges of high dimensionality and sparsity regarding disease development and comorbidity. Chronic diseases have a high likelihood to co-occur, such that patients suffering from one chronic disease have an elevated risk for other diseases in the course of aging. Despite rich results of risk assessment models for chronic diseases, risk prediction considering the complex mechanisms of disease development and comorbidity remains to be underresearched. This research aims to develop tensor factorization-based methods to predict the onset of new chronic diseases through incorporating the comorbidity patterns with the clinical and sequential factors revealed in the electronic health records (EHRs). The efficacy of the proposed methods was validated through predicting the onset of new chronic diseases using the EHRs data for 23 years from a major hospital in Hong Kong. The proposed methods could inform proactive health management programs for at-risk patients with different chronic conditions.