Managing Medical Data to Save Lives
Having a patient’s complete medical history is an important step in providing proper care. Incomplete records, and uncertainties in the type and amount of medications a person uses, can lead to ineffective or even dangerous treatments. In the United States alone, adverse drug events are responsible for nearly 700,000 emergency room visits, 7,000 deaths and more than $20 billion in wasteful medical spending every year. Research has shown that half of such incidents could be preventable with more complete knowledge about patients’ medical history and drug usage. This is why Dr. Ju Long, chair of the Texas State Department of Computer Information Systems and Quantitative Methods, is working to develop a tool that will allow nurses and patients to more easily and accurately collect all relevant medical information in one place.
Long’s background is in economics; she took an interest in social work after noting a disparity in social support in healthcare in her home country of China. She came to the United States to learn about our more robust social work programs and began a career in technology asking “how can you use these new technologies to help people?” She began teaching and conducting research for the McCoy College of Business Administration in 2004.
“Healthcare is really the last frontier of the IT revolution,” explains Long. “Many records are still kept with pen and paper, a nurse asking questions during a preliminary examination and recording the answers.”
“Healthcare is really the last frontier of the IT revolution,” explains Long. “Many records are still kept with pen and paper, a nurse asking questions during a preliminary examination and recording the answers.” This can lead to mistakes in reporting when a nurse records something incorrectly or a patient forgets a medication or doesn’t think it is important to mention. Patients can also be mistaken about the dosage of a medication they’ve taken, especially for over-the-counter meds like ibuprofen. Long and coresearchers have developed a tool that they describe as “an accurate, computationally efficient and scalable algorithm to construct a medical history timeline.” Such a tool “could be integrated into electronic medical records, enabling informed clinical decision-making at the point of care.”
The team demonstrated the effectiveness of their tool during a National Institutes of Health competition in partnership with SureScript, a pharmaceutical information technology firm dedicated to using information technology to “increase patient safety, lower costs and improve quality of care.” Long’s algorithm proved extremely effective at sorting and cataloguing a database of one million prescription records provided by SureScript. The application works by loading a patient’s existing prescription list; using natural language processing to break each prescription record into useful data elements such as the medication name, formulation, frequency and dose; and then generating a visual timeline that a caregiver can review with the patient for accuracy. The application showed an overall success rate of 89.4 percent at accurately identifying the type, dosage and amount of medications being taken by the subject. Further improvements to the model should push the success rate to 95 percent or more.
The next steps in the project include getting doctors and patients involved in further testing and developing an application that patients can use to self-report their medication history in the waiting room. These reports can be cross-examined with existing records and then confirmed by nurses and doctors throughout the treatment process to ensure accuracy, efficiency and ultimately an improvement in patient safety.