Lighting Up Missing Links by Generative AI: A Case of Retinal Detachment

Abstract

This paper proposes a generative-AI-assisted framework for utilizing 1st-person patient narratives to improve medical understanding and decision-making, particularly in the pre-diagnostic or subclinical stages. Contemporary medicine often relies on symptom confirmation, leading to delayed intervention and a heavy cognitive burden on physicians when patients present with fragmented or contradictory accounts. Using the author’s personal experience with acute retinal detachment, this study illustrates how rare, fleeting visual phenomena, such as the preoperative perception of multicolored elliptical particles and postoperative transformation into geometric and network-like patterns, are typically dismissed or lost despite their potential informational value. This paper argues that such narratives, when structured and interpreted through generative AI, can be mapped onto disease progression scenarios and linked with medical knowledge, enabling “structural explanations” rather than passive watchful waiting. This approach, termed N1-Diagnoser, connects 1st-person volatile narratives with large-scale data, literature, and temporal relationships to highlight and clarify hidden events behind the missing links in understanding. These benefits include earlier patient reassurance, im-proved self-observation, reduced physician workload, and enhanced lifestyle guidance. More broadly, this study frames generative AI as a tool for lighting up missing links to enable novel human-centered decision-making via knowledge creation.