Metabolic Profiles for Predicting HIV Immunological Non-Response

101 plasma metabolite and cytokine profiles from people living with HIV, collected at Guangzhou Eighth People’s Hospital in late 2023. The data includes 189 quantified metabolites and 92 cytokines, used to identify six metabolites associated with immunological non-response. Authored by Heping Zhao and published under CC-BY-4.0 in May 2026.

Use Cases

• Train classification models to predict immunological non-response based on the six differential metabolites identified.
• Analyze correlations between specific metabolites like guanidoacetic acid and changes in CD4+ T cell count.
• Explore feature importance for machine learning models, using 2-methylpentanoic acid and guanidoacetic acid as top-ranked features.
• Investigate metabolic dysbiosis as a potential mechanism for poor immune recovery in HIV patients on antiretroviral therapy.

Strengths

• Includes 189 quantified plasma metabolites and 92 cytokines from a cohort of 101 patients.
• Classification models based on six metabolites achieved an AUC of 0.717.
• Data is associated with a specific clinical cohort and detailed methodology (UPLC-MS, Olink PEA).

Limitations

• Row count and column-level documentation are unknown, requiring manual inspection after download.
• The dataset is small at 178.2 KB, limiting its scope for large-scale machine learning.
• Data may reflect geographic bias inherent to a single hospital cohort in Guangzhou.

Provenance

Source

figshare

Collection Method

Plasma metabolites quantified via ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS); cytokines measured via Olink multiplex proximity extension assay (PEA).

Time Range

Data collected November-December 2023.

Freshness

Last updated 2026-05-15 05:44:17.

Geography

Cohort established at Guangzhou Eighth People’s Hospital, China.