Cohort-Derived Machine Learning Models for Individual Prediction of Chronic Kidney Disease in People Living With Human Immunodeficiency Virus: A Prospective Multicenter Cohort Study.
Ontology highlight
ABSTRACT: Background
It is unclear whether data-driven machine learning models, which are trained on large epidemiological cohorts, may improve prediction of comorbidities in people living with human immunodeficiency virus (HIV).Methods
In this proof-of-concept study, we included people living with HIV in the prospective Swiss HIV Cohort Study with a first estimated glomerular filtration rate (eGFR) >60 mL/minute/1.73 m2 after 1 January 2002. Our primary outcome was chronic kidney disease (CKD)-defined as confirmed decrease in eGFR ≤60 mL/minute/1.73 m2 over 3 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/minute/1.73 m2), 1192 (9%) developed a CKD after a median of 8 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.
SUBMITTER: Roth JA
PROVIDER: S-EPMC8514185 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Background Offering living kidney donation raised the concern that donors are exposed to unknown risks. All Swiss transplant centres therefore decided to start a prospective cohort study of living kidney donors in Switzerland. This paper describes the rationale for and implementation of this cohort study. Methods/design All kidney donors in Switzerland are registered and examined before donation and biennially after donation starting in the first year after nephrectomy. Before each follow-up visit, the study centre sends a package to the kidney donor containing the health questionnaire, blood and urine tubes and a prepaid envelope for sending the samples to the central laboratory. The donor makes an appointment with their family physician, who examines the donor and reports findings such as pain and other complaints, blood pressure, creatinine, albumin, all major health events and the state of mental and social well-being to the study centre. The family doctor draws the blood sample and mails it with the urine sample in the prepaid envelope. All data are centrally managed. All abnormal findings in the follow-up of individual donors are regularly discussed with the principal investigator, and necessary clinical changes made and recorded in the database. The health insurance of the recipient covers all costs of the donor follow-up. The main outcomes are the occurrence of albuminuria, hypertension and renal insufficiency. The secondary outcomes are major somatic and social events such as death, cardiovascular disease, stroke and depression. Discussion This prospective cohort offers unique opportunities to assess the risks of living kidney donation and will allow us to examine the risks associated with the methods used for nephrectomy in Switzerland (various forms of open surgery and laparoscopic nephrectomy). The prospective collection of all clinically relevant data and the regular monitoring of donors will allow timely interventions at early stages before serious kidney and general health problems occur.
| S-EPMC3208897 | biostudies-other
Project description:Background: Age-related comorbidities remain a burden in People living with HIV (PLWH). DNA methylation may be a link between HIV, aging, and the increased risk of lung comorbidities. Here, we investigate whether bronchoalveolar lavage (BAL) cells of PLWH demonstrate epigenome wide disruptions and advanced epigenetic aging. Methods: BAL cell DNA methylation profiles from 25 PLWH and 16 uninfected controls were tested for differential methylation of transposable elements Alu and LINE-1, a marker of aging. We used a weighted gene correlation network analysis to identify co-methylation networks associated with HIV and age. We tested the effect of HIV on DNA methylation using a robust linear model. An enrichment analysis was conducted to identify differentially methylated pathways (false discovery rate <0.10). Results: The BAL cells of PLWH were marked by global hypomethylation in both Alu and LINE-1 elements. Six co-methylated CpG networks were identified that were significantly associated with age; of these, the red module was significantly differentially methylated in PLWH and enriched pathways associated with HIV, Ras signaling, T cell receptors, and bacterial invasion of epithelial cells. We identified 6,428 differentially methylated sites associated with HIV. Conclusions: We have shown here for the first time that alterations in the DNA methylation of BAL cells (mainly macrophages) in the lung HIV show in a pattern of advanced aging. In conjunction with our findings in blood cells and airway cells, this study strongly support that HIV may contribute to increase the risk of lung comorbidities through the epigenetic regulation of aging.
2024-07-03 | GSE262656 | GEO
Project description:BackgroundPeople living with HIV might have a poor or delayed response to vaccines, mainly when CD4 cell counts are low, and data concerning COVID-19 vaccines in this population are scarce. This prospective cohort study assessed the safety and immunogenicity of the inactivated SARS-CoV-2 vaccine CoronaVac in people with HIV compared with people with no known immunosuppression.MethodsIn this prospective cohort study, adults (aged ≥18 years) living with HIV who were regularly followed up at the University of Sao Paulo HIV/AIDS outpatient clinic in Sao Paulo, Brazil, were included in the study. Eligibility for people with HIV was independent of antiretroviral use, HIV viral load, or CD4 cell count. Adults with no known immunosuppression with CoronaVac vaccination history were included as a control group. CoronaVac was given intramuscularly in a two-dose regimen, 28 days apart. Blood was collected before vaccine administration and 6 weeks after the second dose (day 69). Immunogenicity was assessed at baseline (day 0), before second vaccine (day 28), and 6 weeks after second vaccine dose (day 69) through SARS-CoV-2 IgG titre and seroconversion, neutralising antibody (NAb) positivity and percentage activity, and factor increase in IgG geometric mean titres (FI-GMT). We investigated whether HIV status and CD4 count (<500 or ≥500 cells per μL) were associated with CoronaVac immunogenicity by use of multivariable models adjusted for age and sex.FindingsBetween Feb 9, 2021, and March 4, 2021, 776 participants were recruited. Of 511 participants included, 215 (42%) were people with HIV and 296 (58%) were people with no known immunosuppression. At 6 weeks after the second vaccine dose (day 69), 185 (91%) of 204 participants with HIV and 265 (97%) of 274 participants with no known immunosuppression had seroconversion (p=0·0055). 143 (71%) of 202 participants with HIV were NAb positive compared with 229 (84%) of 274 participants with no known immunosuppression (p=0·0008). Median IgG titres were 48·7 AU/mL (IQR 26·6-88·2) in people with HIV compared with 75·2 AU/mL (50·3-112·0) in people with no known immunosuppression (p<0·0001); and median NAb activity was 46·2% (26·9-69·7) compared with 60·8% (39·8-79·9; p<0·0001). In people with HIV who had CD4 counts less than 500 cells per μL seroconversion rates, NAb positivity, and NAb activity were lower than in those with CD4 counts of at least 500 cells per μL. In multivariable models for seroconversion, NAb positivity, IgG concentration, and NAb activity after a complete two-dose regimen, adjusted for age and sex, people with HIV who had CD4 counts of at least 500 cells per μL and people with no known immunosuppression had higher immunogenicity than did people with HIV with CD4 counts less than 500 cells per μL. No serious adverse reactions were reported during the study.InterpretationImmunogenicity following CoronaVac in people with HIV seems strong but reduced compared with people with no known immunosuppression. Our findings highlight the need for strategies to improve vaccine immunogenicity in people with HIV.FundingFundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP), Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), and B3-Bolsa de Valores do Brasil.
| S-EPMC8942475 | biostudies-literature
Project description:BackgroundAlthough fever is considered a sign of infection, many individuals with primary immunodeficiency (PI) anecdotally report a lower-than-normal average body temperature on online forums sponsored by the Immune Deficiency Foundation (IDF). There is limited knowledge about the average body temperature and fever response in PI.ObjectiveThis study aims to compare median body temperatures between adults with and without PI diagnoses living in the same household and to engage individuals living with PI throughout the research process.MethodsPatients with PI designed and launched a prospective cohort comparison study as citizen scientists. A multidisciplinary team designed and implemented a patient-informed study with continuous patient-driven input. Median body temperatures were compared between the 2 cohorts using the Mann-Whitney test with Bonferroni correction. The IDF conducted a post-study patient experience survey.ResultsData from 254 households were analyzed (254/350, 72.6% participation rate). The PI population was predominantly female (218/254, 85.8%), White (248/254, 97.6%), and with a median age of 49 years. The non-PI population was largely male (170/254, 66.9%), White (236/254, 92.9%), and with a median age of 53 years. Common variable immunodeficiency was the most common PI diagnosis (190/254, 74.8%). Of the 254 individuals with PI, 123 (48.4%) reported a lower-than-normal nonsick body temperature, whereas 108 (42.5%) reported a normal (between 97°F and 99°F) nonsick body temperature. Among individuals with PI, when infected, 67.7% (172/254) reported the absence of fever, whereas 19.7% (50/254) reported a normal fever response. The recorded median body temperature was minimally but statistically significantly higher for patients with PI in the morning. Although 22.4% (57/254) of patients with PI self-reported illness, a fever of 100.4°F or higher was uncommon; 77.2% (196/254) had a normal temperature (between 97°F and 99°F), and 16.2% (41/254) had a lower-than-normal temperature (between 95.0°F and 96.9°F) when sick. For these sick patients with PI, the median body temperature was minimally but statistically significantly higher for patients in the morning and early evening. Overall, 90.9% (231/254) of participants would be very likely to participate in future IDF studies, although 94.1% (239/254) participants had never taken part in previous studies.ConclusionsTo our knowledge, this is the first study to evaluate average body temperature in individuals with PI. Although there were small statistically significant differences in body temperatures between PI and non-PI subjects, the clinical significance is unclear and should be interpreted with caution, given the methodological issues associated with our small convenience sample and study design. As PIs are heterogeneous, more research is needed about how the fever response differs among diverse PIs compared with healthy controls. This study highlights that individuals with PI are knowledgeable about their health and can offer unique insights and direction to researchers and clinicians.
| S-EPMC7735893 | biostudies-literature
Project description:We aimed to develop machine learning models to accurately predict bronchiolitis severity, and to compare their predictive performance with a conventional scoring (reference) model. In a 17-center prospective study of infants (aged < 1 year) hospitalized for bronchiolitis, by using routinely-available pre-hospitalization data as predictors, we developed four machine learning models: Lasso regression, elastic net regression, random forest, and gradient boosted decision tree. We compared their predictive performance-e.g., area-under-the-curve (AUC), sensitivity, specificity, and net benefit (decision curves)-using a cross-validation method, with that of the reference model. The outcomes were positive pressure ventilation use and intensive treatment (admission to intensive care unit and/or positive pressure ventilation use). Of 1,016 infants, 5.4% underwent positive pressure ventilation and 16.0% had intensive treatment. For the positive pressure ventilation outcome, machine learning models outperformed reference model (e.g., AUC 0.88 [95% CI 0.84-0.93] in gradient boosted decision tree vs 0.62 [95% CI 0.53-0.70] in reference model), with higher sensitivity (0.89 [95% CI 0.80-0.96] vs. 0.62 [95% CI 0.49-0.75]) and specificity (0.77 [95% CI 0.75-0.80] vs. 0.57 [95% CI 0.54-0.60]). The machine learning models also achieved a greater net benefit over ranges of clinical thresholds. Machine learning models consistently demonstrated a superior ability to predict acute severity and achieved greater net benefit.
| S-EPMC7335203 | biostudies-literature
Project description:People with type 2 diabetes mellitus have increased risk of chronic kidney disease and atherosclerotic cardiovascular disease. Improved care delivery and implementation of guideline-directed medical therapy have contributed to the declining incidence of atherosclerotic cardiovascular disease in high-income countries. By contrast, the global incidence of chronic kidney disease and associated mortality is either plateaued or increased, leading to escalating direct and indirect medical costs. Given limited resources, better risk stratification approaches to identify people at risk of rapid progression to end-stage kidney disease can reduce therapeutic inertia, facilitate timely interventions and identify the need for early nephrologist referral. Among people with chronic kidney disease G3a and beyond, the kidney failure risk equations (KFRE) have been externally validated and outperformed other risk prediction models. The KFRE can also guide the timing of preparation for kidney replacement therapy with improved healthcare resources planning and may prevent multiple complications and premature mortality among people with chronic kidney disease with and without type 2 diabetes mellitus. The present review summarizes the evidence of KFRE to date and call for future research to validate and evaluate its impact on cardiovascular and mortality outcomes, as well as healthcare resource utilization in multiethnic populations and different healthcare settings.
| S-EPMC10995482 | biostudies-literature
Project description:BackgroundThe limited available data suggest that the prevalence of problem gambling is increased among young adults with first-episode psychosis, possibly due in part to several risk factors for problem gambling that are common in this population. Aripiprazole, a widely used antipsychotic drug, has also been linked to cases of problem gambling, but causality remains uncertain. Although the consequences of problem gambling further hinder the recovery of people with first-episode psychosis, there is a paucity of research about this comorbidity and its risk factors. Additionally, to our knowledge, no screening instrument for problem gambling tailored to these individuals exists, contributing to its under-recognition. Further, treatment approaches for problem gambling adapted to this population are at an embryonic stage, while existing treatments effectiveness remains to be documented. Using an innovative screening and assessment procedure for problem gambling, this study aims to identify risk factors for problem gambling among people with first-episode psychosis and to document the effectiveness of standard treatment approaches.MethodsThis is a multicenter prospective cohort study conducted in two first-episode psychosis clinics, including all patients admitted between November 1st, 2019, and November 1st, 2023, followed for up to 3 years until May 1st, 2024. These 2 clinics admit approximately 200 patients annually, for an expected sample size of 800 individuals. The primary outcome is the occurrence of a DSM-5 diagnosis of gambling disorder. All patients are screened and evaluated for problem gambling using a systematic procedure at admission, and every 6 months thereafter. Socio-demographic and clinical variables are prospectively extracted from the patients' medical records. The nature and effectiveness of treatments for problem gambling offered to affected individuals are also documented from medical records. Survival analyses with Cox regression models will be used to identify potential risk factors for problem gambling. Descriptive statistics will document the effectiveness of treatments for problem gambling in this population.DiscussionA better understanding of potential risk factors for problem gambling among people with first-episode psychosis will allow for better prevention and detection of this neglected comorbidity. Results of this study will also hopefully raise clinicians' and researchers' awareness and serve as the basis to adapted treatments that will better support recovery.Trial registrationClinicalTrials.gov, NCT05686772. Retrospectively registered, 9 January 2023.
| S-EPMC10127497 | biostudies-literature
Project description:To heighten the awareness of kidney malignancy in patients with HIV infection to facilitate the early diagnosis of kidney cancer, the differentially expressed mRNAs were analyzed in this malignant tumor using RNA sequencing (RNA-seq). We identified 2962 protein-coding transcripts in HIV-associated kidney cancer. KISS1R, CAIX, and NPTX2 mRNA expression levels were specifically increased in HIV-associated kidney cancer, and UMOD and TMEM213 mRNA were decreased in most cases based on real-time PCR analyses. These findings were similar to those noted for the general population with renal cell carcinoma. Immunohistochemical staining analysis also showed that a total of 16 of 18 malignant kidney cases among PLWH exhibited positive staining for KISS1R and CAIX. Pathway analysis of the differentially expressed mRNAs in HIV-associated kidney cancer revealed that several key pathways were involved, including vascular endothelial growth factor-activated receptor activity, IgG binding, and lipopolysaccharide receptor activity. Altogether, our findings reveal the identified molecular changes in kidney malignancy, which may offer a helpful explanation for cancer progression and open up new therapeutic avenues that may decrease mortality after a cancer diagnosis among PLWH.
2022-06-05 | GSE205204 | GEO
Project description:BackgroundRelations between different measures of human immunodeficiency virus-related immunosuppression and chronic kidney disease (CKD) remain unknown.MethodsImmunosuppression measures included baseline, current, time-lagged and nadir CD4, years and percentage of follow-up (%FU) with CD4 ≤200 cells/μL, and CD4 recovery. CKD was defined as confirmed estimated glomerular filtration rate <60 mL/minute/1.73 m2.ResultsOf 33 791 persons, 2226 developed CKD. Univariably, all immunosuppression measures predicted CKD. Multivariably, the strongest predictor was %FU CD4 ≤200 cells/μL (0 vs >25%; incidence rate ratio [IRR], 0.77 [95% confidence interval [CI], .68-.88]), with highest effect in those at low D:A:D CKD risk (IRR, 0.45 [95% CI, .24-.80]) vs 0.80 [95% CI, .70-.93]).ConclusionsLonger immunosuppression duration most strongly predicts CKD and affects persons at low CKD risk more.
| S-EPMC7904289 | biostudies-literature
