Prediction of diabetic kidney disease risk using machine learning models: a population-based cohort study of Asian adults
- Singapore Eye Research Institute, Singapore
- National University of Singapore, Singapore
- Singapore General Hospital, Singapore
Abstract
Background: Machine learning (ML) techniques improve disease prediction by identifying the most relevant features in multi-dimensional data. We compared the accuracy of ML algorithms for predicting incident diabetic kidney disease (DKD).
Methods: We utilized longitudinal data from 1365 Chinese, Malay and Indian participants aged 40-80 years with diabetes but free of DKD who participated in the baseline and 6-year follow-up visit of the Singapore Epidemiology of Eye Diseases Study (2004-2017). Incident DKD (11.9%) was defined as an estimated glomerular filtration rate (eGFR) <60 mL/min/1.73m2 with at least 25% decrease in eGFR at follow-up from baseline. 339 features including participant characteristics, retinal imaging, genetic and blood metabolites were used as predictors. Performances of several ML models were compared to each other and to logic regression (LR) model based on established features of DKD (age, sex, ethnicity, duration of diabetes, systolic blood pressure, HbA1c, and body mass index) using area under the receiver operating characteristic curve (AUC).
Results: ML model, Elastic Net (EN) had the best AUC (95% confidence interval) of 0.851 (0.847-0.856), which was 7.0% relatively higher than by LR 0.795 (0.790-0.801). Sensitivity and specificity of EN were 88.2% and 65.9% vs. 73.0% and 72.8% by LR. The top-15 predictors included age, ethnicity, antidiabetic medication, hypertension, diabetic retinopathy, systolic blood pressure, HbA1c, eGFR and metabolites related to lipids, lipoproteins, fatty acids and ketone bodies.
Conclusions: Our results showed ML together with feature selection improves prediction accuracy of DKD risk in an asymptomatic stable population and identifies novel risk factors including metabolites.
Funding: This study was supported by the National Medical Research Council, NMRC/OFLCG/001/2017 and NMRC/HCSAINV/MOH-001019-00. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Data availability
The authors declare that the data supporting the findings of this study are available within the article and its supplementary materials. The individual participant-level data cannot be shared publicly due to privacy and confidentiality concerns. However, de-identified data may be available upon request with a project proposal to qualified individuals, subject to approval by the Singapore Eye Research Institute and access will require a data sharing agreement. Interested researchers can send data access requests to Prof. Ching-Yu Cheng at the Singapore Eye Research Institute using the following email address: cheng.ching.yu@seri.com.sgProcessed version of the datasets are provided in Supplementary Tables S1-S5.
Article and author information
Author details
Charumathi Sabanayagam
Funding
National Medical Research Council (NMRC/OFLCG/001/2017)
- Gavin Tan
National Medical Research Council (NMRC/HCSAINV/MOH-001019-00)
- Charumathi Sabanayagam
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Reviewing Editor
- Girish N Nadkarni, Icahn School of Medicine at Mount Sinai, United States
Ethics
Human subjects: SEED was conducted in accordance with the Declaration of Helsinki and was approved by the SingHealth Centralised Institutional Review Board [2018/2717, 2018/ 2921, 2012/487/A, 2015/2279, 2018/2006, 2018/2594, 2018/2570]. Informed consent was obtained from all participants.
Version history
- Received: July 14, 2022
- Preprint posted: August 18, 2022 (view preprint)
- Accepted: September 12, 2023
- Accepted Manuscript published: September 14, 2023 (version 1)
- Version of Record published: September 27, 2023 (version 2)
- Version of Record updated: January 4, 2024 (version 3)
Copyright
© 2023, Sabanayagam et al.
This article is distributed under the terms of the Creative Commons Attribution License permitting unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 909
- views
-
- 199
- downloads
-
- 1
- citations
Views, downloads and citations are aggregated across all versions of this paper published by eLife.
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Prediction of diabetic kidney disease risk using machine learning models: a population-based cohort study of Asian adults
eLife 12:e81878.
https://doi.org/10.7554/eLife.81878
Further reading
-
- Cancer Biology
- Epidemiology and Global Health
Background:
Age is the most important risk factor for cancer, but aging rates are heterogeneous across individuals. We explored a new measure of aging-Phenotypic Age (PhenoAge)-in the risk prediction of site-specific and overall cancer.
Methods:
Using Cox regression models, we examined the association of Phenotypic Age Acceleration (PhenoAgeAccel) with cancer incidence by genetic risk group among 374,463 participants from the UK Biobank. We generated PhenoAge using chronological age and nine biomarkers, PhenoAgeAccel after subtracting the effect of chronological age by regression residual, and an incidence-weighted overall cancer polygenic risk score (CPRS) based on 20 cancer site-specific polygenic risk scores (PRSs).
Results:
Compared with biologically younger participants, those older had a significantly higher risk of overall cancer, with hazard ratios (HRs) of 1.22 (95% confidence interval, 1.18–1.27) in men, and 1.26 (1.22–1.31) in women, respectively. A joint effect of genetic risk and PhenoAgeAccel was observed on overall cancer risk, with HRs of 2.29 (2.10–2.51) for men and 1.94 (1.78–2.11) for women with high genetic risk and older PhenoAge compared with those with low genetic risk and younger PhenoAge. PhenoAgeAccel was negatively associated with the number of healthy lifestyle factors (Beta = –1.01 in men, p<0.001; Beta = –0.98 in women, p<0.001).
Conclusions:
Within and across genetic risk groups, older PhenoAge was consistently related to an increased risk of incident cancer with adjustment for chronological age and the aging process could be retarded by adherence to a healthy lifestyle.
Funding:
This work was supported by the National Natural Science Foundation of China (82230110, 82125033, 82388102 to GJ; 82273714 to MZ); and the Excellent Youth Foundation of Jiangsu Province (BK20220100 to MZ).
-
- Epidemiology and Global Health
Background:
Circulating omega-3 and omega-6 polyunsaturated fatty acids (PUFAs) have been associated with various chronic diseases and mortality, but results are conflicting. Few studies examined the role of omega-6/omega-3 ratio in mortality.
Methods:
We investigated plasma omega-3 and omega-6 PUFAs and their ratio in relation to all-cause and cause-specific mortality in a large prospective cohort, the UK Biobank. Of 85,425 participants who had complete information on circulating PUFAs, 6461 died during follow-up, including 2794 from cancer and 1668 from cardiovascular disease (CVD). Associations were estimated by multivariable Cox proportional hazards regression with adjustment for relevant risk factors.
Results:
Risk for all three mortality outcomes increased as the ratio of omega-6/omega-3 PUFAs increased (all Ptrend <0.05). Comparing the highest to the lowest quintiles, individuals had 26% (95% CI, 15–38%) higher total mortality, 14% (95% CI, 0–31%) higher cancer mortality, and 31% (95% CI, 10–55%) higher CVD mortality. Moreover, omega-3 and omega-6 PUFAs in plasma were all inversely associated with all-cause, cancer, and CVD mortality, with omega-3 showing stronger effects.
Conclusions:
Using a population-based cohort in UK Biobank, our study revealed a strong association between the ratio of circulating omega-6/omega-3 PUFAs and the risk of all-cause, cancer, and CVD mortality.
Funding:
Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institute of Health under the award number R35GM143060 (KY). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
