P058
When Bugs Spill Secrets: Leveraging Bacterial Genomes and Machine Learning to Identify Drivers of Recent Tuberculosis Transmission in Accra, Ghana
M N Séraphin(1) M Asare-Baah(1) M A Omari(3) J Chariker(2) J S Afriyie-Mensah(3)
1:University of Florida; 2:University of Louisville; 3:University of Ghana
Tuberculosis (TB) prevalence in Ghana is estimated at 356 cases per 100,000 people, and efforts to improve case detection are ongoing. Understanding the role of high-risk behaviors in recent TB transmission can help refine intervention strategies. From June 2022 to June 2023, 150 new and previously treated pulmonary TB patients were recruited from a tertiary hospital in the Greater Accra Region. Recent transmission was defined as a bacterial genetic distance of ≤12 single nucleotide polymorphisms (SNPs). Bacterial genomes (n = 296) from 96 culture-confirmed cases were analyzed using a supervised learning algorithm to estimate transmission probabilities and individual reproductive numbers (Ri) were calculated. A gamma regression model with a log link was used to assess how social, clinical, and demographic factors were multiplicatively associated with Ri, and model-based population attributable fractions (PAFs) were estimated. The cohort was predominantly male (68%), with 51% in unskilled labor and 15% co-infected with HIV. Unskilled labor was associated with a 1.58-fold increase in Ri compared to skilled labor (p = 0.014). HIV co-infection was associated with a 1.67-fold increase in Ri (p = 0.045). In contrast, older age and prior TB treatment were associated with 35% and 50% reductions in Ri, respectively (p = 0.026 and p = 0.018). The estimated PAF showed that unskilled labor contributed 23.7% (95% CI, 4.1%–41.6%) of recent transmission potential. These findings suggest that occupational and clinical factors significantly influence TB transmission dynamics and should be considered in targeted control strategies.
