Automated Categorization of Systemic Disease and Duration From Electronic Medical Record System Data Using Finite-State Machine Modeling: Prospective Validation Study

• DOI: 10.2196/24490
• Publication Date: 2020-12-17T14:23:04Z
• Authors: Gumpili Sai Prash...; Ayush Deva; Ranganath Vadapalli; Anthony Vipin Das
• Keywords: Medical record; Unstructured data; Medical History; Electronic medical record

Background One of the major challenges in health care sector is that approximately 80% generated data remains unstructured and unused. Since it difficult to handle from electronic medical record systems, tends be neglected for analyses most hospitals centers. Therefore, there a need analyze big systems so we can optimally utilize unearth all unexploited information it. Objective In this study, aimed extract list diseases associated keywords along with corresponding time durations an indigenously developed system describe possibility analytics acquired datasets. Methods We propose novel, finite-state machine sequentially detect cluster disease names patients’ history. defined 3 states transition matrix, which depend on identified keyword. addition, also state-change action essentially each transition. The dataset used study was obtained called eyeSmart implemented across large, multitier ophthalmology network India. included past history contained records 10,000 distinct patients. Results extracted by using accuracy 95%, sensitivity 94.9%, positive predictive value 100%. For extraction duration disease, machine’s 93%, 92.9%, Conclusions demonstrated accurately identify names, keywords, large cohort patient system.