Table 2_Identification of active sequences in HIV-1 molecular transmission networks using machine learning: a cross-sectional study in Hainan Province, south China.docx

Understanding transmission dynamics within molecular networks is critical for precision prevention.</p>Methods<p>We conducted a cross-sectional study of participants living with HIV-1 in Hainan Province between 2021 and 2025. HIV-1 molecular transmission networks were constructed using HIV-TRACE.

Sequences from individuals diagnosed by the end of 2023 were used to construct the baseline network, and sequences from newly reported cases during 2024–2025 were added to form the final network. Active sequences were defined as those from newly reported cases in 2024–2025 or those directly linked to newly reported cases within the final network. Multivariable logistic regression was used to identify factors independently associated with active sequences, and gradient boosting to assess variable importance.</p>Results<p>Among 2,342 sequences, 1,071 (45.7%) were included in the final molecular transmission network, comprising 505 (47.2%) active sequences.