For about 10 years, I practiced as a neurologist with the aim of becoming a general neurologist who can address both psychiatric and neurological symptoms. In the last three of those years, I focused specifically on difficult-to-diagnose cases. During this period, I succeeded in diagnosing approximately 70% of previously undiagnosed patients. Through that experience, I learned that the most crucial factors for making a diagnosis were: conducting an exhaustive review of the patient’s medical history and physical findings, integrating all available information, carefully reading relevant research papers, extracting the key problems, and planning the tests and treatments that could lead to a diagnosis.

In the course of tackling these undiagnosed cases, I realized that considering all potential differential diagnoses—including rare and intractable diseases, as well as atypical presentations—places a significant burden on clinicians, especially when searching through existing literature. Furthermore, the remaining 30% of patients whom I could not diagnose fell into two categories: those with novel genetic mutations lacking a family history, and those with autoantibodies found in blood or cerebrospinal fluid. Recognizing the need for organ- or cell-specific biomarkers, I decided to study abroad to conduct research on extracellular vesicles (EVs). Currently, drawing on these experiences, I am working on:

Investigating inter-organ crosstalk mediated by extracellular vesicles in body fluids and developing new biomarkers.

Leveraging recent advances in artificial intelligence (AI), which have dramatically lowered the cost of learning, to facilitate the utilization of clinical text—particularly in the context of rare and intractable diseases. This includes developing methods to employ ontologies when converting clinical text into structured data, creating new ways to visualize patient data, and thereby reducing communication barriers between patients, their families, and healthcare providers to ensure smoother information sharing.

Additionally, in line with the increasing “commoditization” of AI, one of my lifelong goals is to expand educational efforts that enable more individuals to harness AI in medical and healthcare fields, while also identifying and exploring challenges that AI alone cannot easily address.

Researchmap