AI-Powered Bayesian Statistics in Biomedicine

Statistics and artificial intelligence (AI) are distinct yet closely interconnected disciplines, each characterized by its own historical roots and methodological approaches. This paper explores their collaborative potential, seeking to answer a pivotal question: How can statistics and AI collaborate to extract valuable insights from complex data? Within this context, we present three compelling case studies that showcase the harmonious integration of statistics and AI for the analysis of high-resolution pathology images, an emerging type of medical image that provides rich cellular-level information and serves as the gold standard for cancer diagnosis. Furthermore, recent advancements in spatial transcriptomics, which typically yield paired digital pathology images from the same tissue sample, introduce a new dimension to pathology images. This evolving landscape extends the horizons of the proposed AI-statistics framework, holding a promise of propelling biomedical research into new territories and delivering breakthroughs in our understanding of complex diseases.