| Title: | Development and evaluation of cardiovascular disease-specific frailty index: a machine learning based analysis of the UK biobank |
| Journal: | European Journal of Preventive Cardiology |
| Published: | 22 Aug 2025 |
| Pubmed: | https://pubmed.ncbi.nlm.nih.gov/40842330/ |
| DOI: | https://doi.org/10.1093/eurjpc/zwaf452 |
Publication 19688
WARNING: the interactive features of this website use CSS3, which your browser does not support. To use the full features of this website, please update your browser.
Abstract
AIMS: Frailty assessments targeting cardiovascular diseases (CVD) risk lack specificity. Our study aims to develop and validate a CVD-specific FI (FICVD) using machine learning algorithms.</p>
METHODS AND RESULTS: 366 622 included participants from the prospective cohort of UK Biobank were split to the development, temporal validation and spatial validation cohorts. Elastic net regression was conducted in the development cohort to obtain the coefficients for each health-related items of frailty index (FI) to construct a CVD-specific FI (FICVD). A polygenic risk score (PRS) for CVD was generated to quantify participants' genetic susceptibility to CVD. Results showed the hazard ratios for a 1 SD increase in FICVD and FI for the risk of incident CVD were 1.40 (95% CI: 1.37-1.43) and 1.26 (95% CI: 1.24-1.27) in the development cohort, respectively. The area under the curve value for FICVD was significantly higher than that for FI in predicting 10-year CVD (0.649 vs. 0.605) in the development cohort. Similar patterns were observed in both the temporal validation and the spatial validation cohort. In addition, participants with high FICVD and high genetic risk have a 2.84-fold (95%CI: 2.79-2.96) risk of CVD, compared with participants with low FICVD and low genetic risk.</p>
CONCLUSION: Newly developed FICVD demonstrated better performance in prediction and risk stratification of CVD. Also, combining FICVD with genetic susceptibility can assist in the identification of high-risk individuals for CVD. Our findings collectively emphasize the novel FICVD can optimize individualized` CVD risk assessment, and provide potential strategies for CVD prediction and early intervention.</p>
LAY SUMMARY: This study conducted based on a prospective cohort and machine learning algorithms to optimize frailty assessment strategies for CVD risk. Key findings in our research are as follows:Construction of FICVD requires fewer health-related items than the construction of traditional FI, which may substantially reduce the burden of data collection.FICVD demonstrated better performance in both prediction and risk stratification of CVD than traditional FI.Combination of FICVD and genetic susceptibility can further enhance the identification of high-risk individuals for CVD.</p>
4 Authors
- Jiatang Xu
- Zhensheng Hu
- Kai Huang
- Zhongkai Wu
1 Application
| Application ID | Title |
|---|---|
| 100739 | Artificial intelligence for using multi-modal data to improve the identification and prediction of diseases in the UK Biobank |
Enabling scientific discoveries that improve human health