Skip to navigation Skip to main content Skip to footer

Approved Research

Machine learning to reveal the natural history and prognostic factors of type II diabetes mellitus: UK Biobank cohort study

Principal Investigator: Professor Charles Tzu-Chi Lee
Approved Research ID: 82629
Approval date: December 14th 2022

Lay summary


Machine learning which is used to learn patterns of complications at which type II diabetes mellitus (T2DM) patients' inpatient admitted record and to identify prognostic factors related to this natural history is rare in literature. The proposed study aims to fill the gap in the literature through an understanding of the patterns and prognostic factors of complications of T2DM.

Scientific rationale:

We will carry out this study using information from the UK Biobank. Details of demography, outpatient care, hospital inpatient care, drug prescription, image analysis, lifestyle, and genomic and other biomarker data are provided by the UK biobank. T2DM will comprise a case group, and the control group with a 1:1 ratio is selected by exact matching for sex and age. The index date is the first diagnosis date of T2DM and assigning to match controls. Study outcomes for patient's complications are annual inpatient diagnosis International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) first 3 digits record.

Project duration:2023/1/1-2025/12/31

Public health impact:

The present study will demonstrate the following important issues for clinical and public health.

  1. Does the T2DM complication natural history exists patterns? If so, how many predominate patterns and in what contexts?
  2. Does any characteristic or biomarker associated with previous predominate patterns?

Health promotion or medical therapy based on these associated biomarkers can be further studied to confirm their efficacy in the progress of T2DM.