# Analysis of Associated Factors and Construction of Risk Prediction Models for Frailty in Hospitalized Older Adults Living With HIV: Protocol for a Prospective Observational Study

## TL;DR

This study aims to develop a risk prediction model for frailty in older adults living with HIV in China to enable early intervention and improve health outcomes.

## Contribution

The study introduces a machine learning-based risk prediction model for frailty in hospitalized older adults living with HIV.

## Key findings

- A prospective observational study is underway to assess frailty in hospitalized older adults living with HIV.
- Three machine learning algorithms will be used to construct and validate a frailty risk prediction model.
- The model's performance will be evaluated using metrics like accuracy, precision, and area under the curve.

## Abstract

The aging trend of people living with HIV or AIDS in China is increasing day by day. Frailty is a common condition among older adults living with HIV or AIDS and represents a significant cause of poor prognosis, including falls, decreased quality of life, increased mortality, and potentially prolonged hospital stays. Consequently, early frailty screening in this population holds important clinical significance.

This study aims to describe the theoretical basis, research objectives, and implementation plan of a prospective observational study. It will focus on investigating the current status of frailty syndrome in hospitalized older adults living with HIV or AIDS, while simultaneously exploring the development of a clinically applicable risk prediction model.

This study is an ongoing single-center prospective observational study, with a plan to recruit at least 556 hospitalized older adults living with HIV or AIDS (n=445 for development and n=111 for validation). According to the theory of unpleasant symptoms, candidate predictors are categorized into physiological factors (including sociodemographic factors, disease-related influencing factors, sleep, nutrition, and neurocognitive function), psychological factors (including anxiety and depression status), and environmental factors (including social support status). Potential predictors are screened using univariate analysis and least absolute shrinkage and selection operator regression to identify variables for final model inclusion. Model construction and validation employ 3 standard machine learning algorithms: logistic regression, random forest, and support vector machine. Model performance will be evaluated by reporting accuracy, precision, sensitivity, specificity, and the area under the curve.

This study is conducted at a designated infectious disease hospital in Changsha, Hunan Province, China. Participant recruitment commenced on December 22, 2024, and as of December 5, 2025, a total of 603 patients have been enrolled. The primary study findings are anticipated to be published in August 2026.

The findings of this study are expected to provide clinicians in the department of infectious diseases with a convenient tool for frailty risk prediction, thereby enabling early intervention and ultimately improving the long-term health status and quality of life of people living with HIV.