# Association rule mining and network analysis of the evolving comorbidity patterns in HIV inpatients in Baise, China

## TL;DR

This study analyzes comorbidity patterns in HIV patients in Baise, China, using data from 2019 to 2024 to track changes and identify key diseases.

## Contribution

The study introduces a novel temporal and network-based analysis of evolving comorbidity patterns in HIV patients using association rule mining and social network analysis.

## Key findings

- 99.48% of HIV inpatients had two or more comorbidities, with a median of 9 comorbidities per patient.
- Comorbidity patterns evolved from (B20 + B37 → B99) in 2019 to (J18 + E87 → E46) by 2023.
- Electrolyte imbalances (E87), HIV-related infections (B20), and candidiasis (B37) were core disease nodes in the network.

## Abstract

With the widespread use of antiretroviral therapy, human immunodeficiency virus (HIV) infection is considered to be a manageable chronic disease, but it is accompanied by an increased burden of comorbidities. Baise is an area characterized by a high incidence of HIV infection in Guangxi, China. However, research on its comorbidity patterns is limited. This study aims to clarify the burden, patterns, network features, and temporal evolution of comorbidities among HIV inpatients in Baise. We collected electronic medical records from 3,294 HIV patients hospitalized at Baise People’s Hospital between January 2019 and August 2024. The Apriori algorithm was employed to extract association rules between diseases, while Gephi was utilized to construct comorbidity social network diagrams of the data. The findings revealed that 99.48% of patients presented with two or more comorbidities, with a median of 9 comorbidities per patient. Notably, the median number of comorbidities peaked at 11–12 during 2021–2022, subsequently decreasing to 7 in 2023–2024. The comorbidity patterns transitioned from (B20 + B37 → B99) in 2019 to (E46 + B20 → E87 + D64) in 2021 and ultimately evolved into (J18 + E87 → E46) by 2023. Social network analysis indicated that electrolyte imbalances (E87), HIV-related infections (B20) and candidiasis (B37) served as the core disease nodes within the network, displaying close connections with numerous other disease nodes. The identified specific comorbidity patterns can serve as early warnings and screening tools in clinical practice and they underscore the necessity for multidisciplinary care for HIV patients.