Application of Large Electronic Medical Database for Detecting Undiagnosed Patients in the General Population

Clinical application of accumulated medical big data is a hot topic in medical informatics. Not only for suggesting possible diagnoses in each individual, large medical database can be possibly used for detecting undiagnosed patients in the general population. In this study, we tried to develop a computerized system of detecting overlooked undiagnosed patients with rare chronic diseases in the community population by utilizing the uniformed national medical insurance record database. A cumulative total of 489,823 hospital visits at one tertiary medical center were collected for this project. As the target disease, we selected esophagogastric junction outflow obstruction (EGJOO), including achalasia, which is known to be easily overlooked without performing a barium swallow test. Patient selection software automatically picked out 17,814 individuals with the given suspected diagnoses that could be misdiagnosed in patients with the target disease, from which the software further picked out 526 individuals who underwent upper endoscopy but did not undergo barium swallow test. Of them, the hospital medical records suggested that 39 people still suffered from prolonged symptoms lasting for more than 6 months after the first hospital visit. Among them, 16 individuals agreed to undergo the barium swallow test. One of them was confirmed to suffer from EGJOO, possibly based on some undiagnosed connective tissue diseases. An automated computerized detection system with uniform big medical data would realize more efficient and less expensive screening system for undiagnosed chronic diseases in the general population based on symptoms and previously performed examinations in each individual.