Cohort-derived machine learning models for individual prediction of chronic kidney disease in people living with HIV: a prospective multicentre cohort study.

Roth, Jan A; Radevski, Gorjan; Marzolini, Catia; Rauch, Andri; Günthard, Huldrych F; Kouyos, Roger D; Fux, Christoph A; Scherrer, Alexandra U; Calmy, Alexandra; Cavassini, Matthias; Kahlert, Christian R; Bernasconi, Enos; Bogojeska, Jasmina; Battegay, Manuel (2021). Cohort-derived machine learning models for individual prediction of chronic kidney disease in people living with HIV: a prospective multicentre cohort study. The journal of infectious diseases, 224(7), pp. 1198-1208. Oxford University Press 10.1093/infdis/jiaa236

Preview

Text Rauch A.pdf - Accepted Version Available under License Creative Commons: Attribution-Noncommercial-No Derivative Works (CC-BY-NC-ND). Download (1MB) | Preview

BACKGROUND

It is unclear whether data-driven machine learning models, which are trained on large epidemiological cohorts, may improve prediction of co-morbidities in people living with HIV.

METHODS

In this proof-of-concept study, we included people living with HIV of the prospective Swiss HIV Cohort Study with a first estimated glomerular filtration rate (eGFR) >60 ml/min/1.73 m2 after January 1, 2002. Our primary outcome was chronic kidney disease (CKD) ─ defined as confirmed decrease in eGFR ≤60 ml/min/1.73 m2 over three months apart. We split the cohort data into a training set (80%), validation set (10%), and test set (10%) ─ stratified for CKD status and follow-up length.

RESULTS

Of 12,761 eligible individuals (median baseline eGFR, 103 ml/min/1.73 m2), 1,192 (9%) developed a CKD after a median of eight years. We used 64 static and 502 time-changing variables: Across prediction horizons and algorithms and in contrast to expert-based standard models, most machine learning models achieved state-of-the-art predictive performances with areas under the receiver operating characteristic curve and precision recall curve ranging from 0.926 to 0.996 and from 0.631 to 0.956, respectively.

CONCLUSIONS

In people living with HIV, we observed state-of-the-art performances in forecasting individual CKD onsets with different machine learning algorithms.

Interest & Impact

Downloads

85 since deposited on 16 Jun 2020
27 in the past 12 months

Citations

5 Citations in Web of Science ®
6 Citations in Scopus

Search

in Google Scholar™

Services

Actions (login required)

Edit item

Actions (login required)

Edit item