Explainable machine learning framework for biomarker discovery by combining biological age and frailty prediction (Preprint)

Published: May 9, 2024

Biological age (BA) and frailty represent two distinct health measures that offer valuable insights into the aging process. Existing machine learning (ML) predictors of BA frailty, derived from blood-based biomarkers, lack capability to compare analyze these biomarkers comprehensively. Such comparisons may provide deeper pathways. Objective: This study aimed develop a framework by combining ML with eXplainable Artificial Intelligence (XAI) techniques. We utilized data middle-aged older Chinese adults (≥45 years) in 2011/2012 wave (n=9702) 2015/2016 (n=9455, as test set validation) China Health Retirement Longitudinal Study (CHARLS). Sixteen were used predict frailty. Four tree-based algorithms employed training validation, performance metrics compared select best models. Then, SHapley Additive exPlanations (SHAP) analysis was conducted on selected CatBoost performed predictor, Gradient Boosting predictor. Traditional feature importance identified cystatin C glycated hemoglobin major contributors for their respective However, subsequent SHAP demonstrated only primary contributor both models, suggesting it plays an important role Our novel integrates XAI techniques biomarkers. The present approach leverages routine blood can easily incorporate additional providing scalable comprehensive toolset offers quantitative understanding interesting complex physiological traits.

Language: Английский

---

Circulating CXCL9, monocyte percentage, albumin, and C-reactive protein as a potential, non-invasive, molecular signature of carotid artery disease in 65+ patients with multimorbidity: a pilot study in Age.It

Frontiers in Endocrinology, Journal Year: 2024, Volume and Issue: 15

Published: July 23, 2024

Background Carotid endarterectomy (CEA) for the prevention of upcoming vascular and cerebral events is necessary in patients with high-grade stenosis (≥70%). In framework Italian National project Age.It, a pilot study was proposed aiming at discovery molecular signature predictive potential carotid comparing 65+ asymptomatic symptomatic inpatients. Methods A total 42 inpatients have been enrolled, including 26 men 16 women, mean age 74 ± 6 years. Sixteen ≥70% underwent CEA, according to recommendations European Society Vascular Surgery Surgeons. Plaque biopsies peripheral blood samples from same individuals were obtained. Hematobiochemical analyses conducted on all inpatients, plasma cytokines/molecules, such as microRNAs (miRs), IL-6, sIL-6Ralpha, sgp130, myostatin (GDF8), follistatin, activin A, CXCL9, FGF21, fibronectin, measured using ELISA standard technique. MiR profiles obtained phase four (both plaque samples), testing 734 miRs. MiRs emerging profiling comparison validated through RT-qPCR analysis cohort. Results conclusion The two groups differ expression levels c-miRs-126–5p -1271–5p (but not their plaques), which are more expressed subjects. Three cytokines significant between groups: GDF8, CXCL9. Using receiver operating characteristic (ROC) machine learning-based approach, most encompasses albumin, C-reactive protein (CRP), percentage monocytes, allowing distinction (AUC = 0.83, 95% c.i. [0.85, 0.81], p 0.0028). will be tested second cohort monitored patients, application model final evaluation cost/benefit an assessable screening test.

Language: Английский

---

Machine Learning Approaches for Predicting Frailty Base on Multimorbidities in US Adults Using NHANES Data (1999-2018)

Computer Methods and Programs in Biomedicine Update, Journal Year: 2024, Volume and Issue: unknown, P. 100164 - 100164

Published: Sept. 1, 2024

Language: Английский

---

Depression and Plasma pTau₁₈₁ Levels Are Associated with Frailty Status in Hispanic Community-Dwelling Older Women

Gerontology and Geriatric Medicine, Journal Year: 2024, Volume and Issue: 10

Published: Jan. 1, 2024

The population of Hispanic older adults is growing along with the burden chronic diseases. This cross-sectional study aims to assess factors associated frailty among community-dwelling women aged ≥60 years ( n = 357) enrolled in Panama Aging Research Initiative—Health Disparities cognitive-functional health persons Panama. Cognitive function was assessed a neuropsychological test battery. Depression measured Geriatric Scale. Frailty defined using Fried criteria and participants were classified as non-frail, pre-frail or frail. A subsample 281) provided fasting blood samples for quantification protein biomarkers. Associations examined hierarchical multiple linear regressions. 59.4% 9.0% M 69.2 years, SD 6.3) frail, respectively. Having more depression (β .28, p < .001) significantly frailty, even after covariate adjustment. not frailty. Higher pTau181 levels increased .13, .039), whereas higher α2M decreased −.16, .004). These findings advance search indicators biomarkers warrant further studies decrease women.

Language: Английский

---

Parishin Alleviates Pulmonary Fibrosis by Reducing CD38 Levels in Naturally Aging Mice

Rejuvenation Research, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 24, 2024

Parishin, a natural compound, has demonstrated significant potential in mitigating age-related phenotypes and improving outcomes age-associated diseases. Given that aging is major risk factor for numerous chronic conditions, including pulmonary fibrosis, we investigated parishin's effects on cellular senescence lung health. In our study, treated mouse epithelial cells with parishin observed reduction markers alongside an upregulation of sirtuin 1 (SIRT1). Building these

Language: Английский

---