|
Knowledge Sharing and Event Alerting for Patient Safety
“To err is human.” Each year, 44,000-98,000 deaths may be due to medical errors, according to a report published by the Institute of Medicine (IOM) in 1999. That figure ranks medical errors anywhere from the fifth to the eighth leading cause of death in the United States. The report estimated costs of medical errors caused by preventable adverse events was approximately between $17 billion and $29 billion annually. For this reason, in the past six years patient safety has drawn a significant amount of attention from local hospitals to the federal government. All over the country adverse event reporting systems have been implemented and used by patient safety officers to understand the root causes of medical errors. Research results from those systems are mainly on the premise of report collection, correlation between report contents and possible medical errors, and some resolutions to reduce potential errors. However, very little effort has been made to utilize the massive amount of adverse reports and electronic medical records to alert safety officers and flag possible medical errors. Moreover, to our knowledge, none of the existing systems are able to provide an electronic environment for patient safety officers to share their experiences and learn the actions others took to prevent potential errors.
This project will leverage what the patient safety community has accomplished by developing computer algorithms that are expected to achieve the following aims:
1. Understanding semantics from structured and unstructured medical reports by developing software that integrates algorithms from the areas of natural language processing (NLP), association rules mining, machine learning, and information retrieval.
2. Linking adverse event reports with electronic medical records by correlating results obtained from Aim 1 with individual patient’s charts, such as routine medication, allergies, personal history, family history, social history, etc. over a specific period of time.
3. Developing a prototype real-time alert system for patient safety officers that will warn before any harm has been done to the patients.
4. Building a case-based reasoning (CBR) system for knowledge repository and sharing for patient safety community by automatically indexing adverse event reports and making the report database a searchable knowledge base.
|
