Project Details

Description

A 2006 report from the Institute of Medicine of the National Academics says preventable medication errors harm 1.5 million people every year, and cost at least $3.5 billion in extra medical costs to treat drug error-related injuries. In other reports, medication errors account for approximately 1 out of 131 outpatient and 5 out of 854 inpatient deaths. Medication errors could occur in variety of medical workflows such as during prescribing, transcribing at the pharmacy, or at medication administration process. Thus, medication errors are common,life threatening, costly, but preventable. Number of studies reported that health information technology and especially clinical decision support systems (CDSS) were the most important role not only to improve patient safety
and reduce “Medication error” or “Near miss” events. But they also are crucial issue to evaluate hospital performance and improve patient outcome.

Moreover, most of CDSS which were implemented for automated methods statistically developed and maintained by experts at a significant cost to maintain promise and evidence, were “Rule based” methods with limited features itself. In this project, we aim to use “Big data” approach and deep neural network (DNN) model to detect and prevent the medication errors in computer physician order entry (CPOE) system. The DNN model will be verified and evaluated by compared to probabilistic model and human experts in order to specify its accuracy. The model will be using the powerful of Taiwan’s National Health Insurance Big data to able detect sensitively the inappropriate medications prescribed in physicians’ realtime practices. In this study, the potential inappropriate rate of prescription is 4.5-5%, we could notify accurately
50-80% inappropriate. Therefore, versus 300 million prescriptions dispended in Taiwan per year, it could be preventable up to 1.5 million inappropriate prescriptions, which appear as the highest medical errors of “prescription open phase”. Thus, it could be an efficient tool for automatic identification the medication errors in a given prescription and aid in improving patient safety
and quality of care.
StatusFinished
Effective start/end date11/1/1610/31/17

Keywords

  • medication errors
  • deep neural network
  • machine learning
  • probabilistic model
  • Taiwan National Health Insurance database

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.