fix abstract alignment.

events/stancon2024/index.md

@@ -64,9 +64,9 @@ We thank our current sponsors and supporting institutions: Daiichi-Sankyo, Metru
 
 These are the confirmed keynote speakers:
 
-* Chris Wymant (Senior Researcher, Pandemic Sciences Institute, Oxford University)
-_Bayesian multilevel causal modelling of the frequency and implications of having two HIV infections_
-**Abstract:** Individuals with a second HIV infection—a state called dual infection—are those with two distinct strains of HIV virus, which could have been acquired sequentially (“superinfection”) or simultaneously (“coinfection”). Many previous studies have analysed the frequency and implications of dual infection in datasets of less than 200 individuals. These studies have provided limited data to answer a question often raised in clinical contexts: does getting infected again worsen prognosis? We investigated this question in the BEEHIVE project, using data from 2693 individuals from seven national European cohorts and Uganda. We developed a Bayesian multilevel causal model to jointly estimate the set of causal effects connecting age, sex, mode of transmission, viral loads, longitudinal measurements of CD4 (immune system) cells, being dually or singly infected, and numbers of viral strains detected in deep-sequencing data along the whole viral genome. We determined the latter using phylogenetic ancestral state reconstruction of which person each virus was in. We allowed for reverse-causal association between two viral strains and viral load due to detection bias. Four types of random effect defined per individual were all analytically marginalised. To account for frequent low-level contamination in deep-sequence data we used a mixture model for signal and noise, imposed thresholds on the number of genetic sequence fragments supporting a second viral strain, and varied these with sensitivity analyses. Results indicate only a small effect of dual infection on viral loads and CD4 cell decline during untreated infection.
+* Chris Wymant (Senior Researcher, Pandemic Sciences Institute, Oxford University)  
+**Bayesian multilevel causal modelling of the frequency and implications of having two HIV infections**  
+**Abstract:** _Individuals with a second HIV infection—a state called dual infection—are those with two distinct strains of HIV virus, which could have been acquired sequentially (“superinfection”) or simultaneously (“coinfection”). Many previous studies have analysed the frequency and implications of dual infection in datasets of less than 200 individuals. These studies have provided limited data to answer a question often raised in clinical contexts: does getting infected again worsen prognosis? We investigated this question in the BEEHIVE project, using data from 2693 individuals from seven national European cohorts and Uganda. We developed a Bayesian multilevel causal model to jointly estimate the set of causal effects connecting age, sex, mode of transmission, viral loads, longitudinal measurements of CD4 (immune system) cells, being dually or singly infected, and numbers of viral strains detected in deep-sequencing data along the whole viral genome. We determined the latter using phylogenetic ancestral state reconstruction of which person each virus was in. We allowed for reverse-causal association between two viral strains and viral load due to detection bias. Four types of random effect defined per individual were all analytically marginalised. To account for frequent low-level contamination in deep-sequence data we used a mixture model for signal and noise, imposed thresholds on the number of genetic sequence fragments supporting a second viral strain, and varied these with sensitivity analyses. Results indicate only a small effect of dual infection on viral loads and CD4 cell decline during untreated infection._
 
 * Mitzi Morris (Stan Developer, Columbia University)
 * Vianey Leos Barajas (Assistant Professor, Department of Statistical Sciences & School of the Environment, University of Toronto)