Oral Presentation Australian and New Zealand Society for Geriatric Medicine Annual Scientific Meeting 2026

Disease-Related & Age-Related Frailty: A Machine Learning Approach to Characterising Frailty By Aetiology (#26)

Trent Payne 1 , Ruth Eleanor Hubbard 1 2 , Emily Gordon 1 2 , Alyra Shaw 1 , Mark Midwinter 3 , Ross Francis 4
  1. Australian Frailty Network, The University of Queensland, Woolloongabba, QLD, Australia
  2. Department of General and Geriatric Medicine, Princess Alexandra Hospital, Metro South Hospital and Health Service, Brisbane
  3. School of Biomedical Sciences, The University of Queensland, Brisbane
  4. Department of Urology and Kidney Transplantation, Princess Alexandra Hospital, Metro South Hospital and Health Service, Brisbane

Background: Frailty is characterised by reduced physiological reserve and associated with ageing. However, younger individuals with chronic diseases like kidney failure (KF) also develop frailty. Whether disease-related and age-related frailty represent distinct conditions clinically and pathophysiologically remains unknown.

Methods: This cross-sectional analysis used baseline data from the Reversing Frailty in Transplantation (ReFIT) study, comprising 61 KF patients (36 transplant recipients, 25 waitlisted dialysis patients) and 50 community-dwelling older adults. Demographic, clinical, and frailty index data were analysed using regression modelling and machine learning (Random Forest, Regularised Discriminant Analysis), identifying distinguishing features between disease-related and age-related frailty.

Results: Despite similar FI scores (KF: 0.23±0.08 vs older adults: 0.23±0.09, p=0.8), groups demonstrated distinct compositional profiles. Older adults exhibited deficits in balance (54% vs 84% intact, p=0.001), continence (58% vs 97% intact, p<0.001), and instrumental activities of daily living (46% vs 95% independent, p<0.001). KF patients had greater medical complexity, with 61% having hyperpolypharmacy versus 18% in older adults (p<0.001). Regression identified condition-specific pathways: in KF, frailty was best explained by medical burden and disease-related factors (data-driven model: comorbidities, sleep, and four additional issues; adjusted R²=0.81, p<0.001), whereas functional impairments predominated in older adults (functional model: balance, elimination, strength, mobility, and nutrition; adjusted R²=0.78, p<0.001). Machine learning achieved 98.2% cross-validated accuracy (κ=0.96), identifying chronic kidney disease, polypharmacy, mobility, and sleep disturbance as discriminators.

Conclusions: Disease-related and age-related frailty appear to represent distinct conditions with differing mechanisms and deficit profiles. Aggregate frailty scores may obscure clinically meaningful heterogeneity, supporting the need for condition-specific assessment.