Speakers

Despite major advances, over 39 million people globally are still living with HIV, and with nearly 1.3 million new infections in 2023 alone, continued research is vital to improve quality of life, fight stigma, and close persistent gaps in prevention and care. HIV prevention efforts require the optimal distribution of resources like pre-exposure prophylaxis (PrEP) to maximize impact, particularly in high-risk populations such as men who have sex with men (MSM). While epidemic control theory suggests prioritizing those at highest risk of exposure, estimating this risk can be difficult in marginalized populations due to stigma and complex social dynamics. This talk presents a network-based framework for evaluating and designing HIV prevention policies, using contact patterns and simulations to optimize PrEP distribution strategies.

We begin by evaluating different targeting strategies for PrEP, focusing on how network properties influence intervention effectiveness. By analyzing contact patterns within high-risk populations, we show that non-selective PrEP distribution may outperform risk-based strategies, offering a simpler and more efficient approach to reducing HIV transmission. Additionally, selective strategies targeting high-degree nodes in network structures can further enhance prevention outcomes, particularly in populations with highly connected subgroups.

Next, we shift to the cost-effectiveness of various PrEP policies. Using agent-based models of HIV transmission, we assess how factors like PrEP coverage, adherence rates, and healthcare costs influence policy success. Our analysis highlights that broad eligibility criteria maximize both effectiveness and cost-effectiveness, especially in resource-constrained settings, while more restrictive policies can limit access and reduce overall impact.

Through these insights, we demonstrate how network-based models can inform both the design and evaluation of more efficient, equitable, and scalable HIV prevention policies.

Research on rare autoimmune and autoinflammatory diseases has revealed key regulators of immune homeostasis through focused, single gene–single disease studies. However, these sequential and noncoordinated efforts have often left the broader picture of complex molecular interactions underexplored leading to a diagnostic and therapeutic gap and unmet medical need. To address this gap, systematic network science approaches are employed to integrate diverse datasets. Yet, these innovative strategies must contend with significant challenges, such as the scarcity of standardized, high-quality data, which has limited our ability to perform comprehensive, integrative analyses.  In this presentation, I will showcase our recent efforts to provide a more comprehensive understanding of rare disease mechanisms and offer deeper insights into their complex phenotypes. In particular I will focus on developing and adopting standardized nomenclatures for diverse data types, leveraging innovative network-based methodologies, and  integrating heterogeneous data sources with state-of-the-art visualization tools.  

There has been extensive research on disease networks. The disease network is a useful tool that can identify disease characteristics and capture progression trajectories. However, previous disease networks were developed based on network science, which limits their applicability to real-world public health purposes. Therefore, our study aims to develop an unbiased disease progression network and explore its potential public health applications. We used national health insurance claims data of approximately 50 million total Korean population, collected from 2011 to 2022. We used only the primary ICD-10 diagnosis codes and re-grouped into 604 disease categories. To develop a more unbiased disease network, disease occurrence associations were calculated by stratifying by age and sex. Then, Inverse Probability Weighting (IPW) analyses were performed with adjusting confounding/mediating diseases according to the DAG, income, region, BMI, alcohol consumption, smoking, exercise, and disability. Relative risk, attributable risk, and transition probabilities were estimated by IPW. We identified statistically significant disease pairs (RR > 1.5, q-value < 0.00001) and repeated the IPW analysis up to the third round because the network would be updated every round. The significant disease pairs identified in the last round were connected to develop the overall network and subnetworks by age and sex. After the third round of IPW analysis, the disease progression network consisted of 74,745 pairs. Based on this network, we identified disease importance, indirect medical costs, a health deterioration model, and a multimorbidity index from a public health perspective. The disease network research represented the overall progression relationship among diseases and could be highly beneficial for public health policymakers and researchers.