NETWORKS OF CAUSAL EVIDENCE: BIOBEHAVIORAL INFLUENCE IN CHRONIC DISEASE PROGRESSION
Understanding the influence of psychological factors on chronic disease progression is crucial for effective treatment and has garnered significant interest in health and medical research. Despite extensive empirical studies, the complex web of causal relationships among various factors often obscures important indirect connections. To address this challenge, we developed an algorithm using OpenAI's GPT-4 model to automate the identification and organization of causal relationships from text into a structured table, facilitating the construction of a visual network. We applied this approach to 14 review articles about factors influencing cancer progression published between 1997 and 2020, utilizing an In-Context Learning (ICL) method. The extracted factors were categorized into biobehavioral, cellular and biological, and survival and tumor progression. The resulting automated network revealed 2073 causal relationships between 1554 factors, with significant overlap in key factors such as depression, stress, and immune response, when compared to a previously constructed manual network. Additionally, our methodology involved an extensive search in relevant scientific databases to gather studies and articles related to biobehavioral factors and disease progression. This comprehensive analysis, leveraging artificial intelligence to extract relevant data, aims to establish causal networks that reveal patterns and trends, exploring the mechanisms linking biobehavioral factors to disease outcomes. By providing a rigorous examination of the impact of these factors, the findings will enhance our understanding of chronic disease progression, guide clinical practice, and inspire interventions to improve patient outcomes and disease mitigation.