New study paves the way for collaboration on artificial intelligence modelling and medication error reduction globally
Researchers at Harvard Medical School, Brigham and Women's Hospital, Taipei Medical University, and Aesop Technology, a Taiwan-based startup, announced today the results of a new joint study into the international transferability of machine learning (ML) models to detect medication errors. The results were recently published in the Journal of Medical Internet Research - Medical Informatics.
Working to Reduce Medication Errors
Medication errors are a growing financial and healthcare burden that results in economic costs of around US$ 20 billion and more than 250,000 deaths annually in the U.S. alone.
Medication errors can occur during any stage of the medication process, including prescribing, dispensing, administration, and monitoring, with errors in prescribing accounting for 50% of the total.
When medicating patients, physicians go through complex decision-making processes to accurately write a prescription. First, they must clearly define the patient's problem and list the therapeutic objective before selecting an appropriate drug therapy based on age, gender, and possible allergies. They must also consider dosing, drug-drug interaction, potential discontinuation of the drug, drug cost, and other therapies — and all of these need to be done instantly and simultaneously.
"Reducing medication errors at the source is crucial. However, to help physicians be better informed and make better decisions, they need more accurate suggestions and alerts. This is where machine learning can help to make better decisions and improve patient safety and quality of care," said Dr. David W. Bates, Chief of General Internal Medicine and Primary Care at Brigham & Women's Hospital and Professor of Medicine at Harvard Medical School.
For technology to assist in solving these problems, it is critical that machine learning understands these variables. For this to be successful, data must be properly collected, organized, and maintained.
Taiwan is one of the world's few countries with a centralized and well-structured electronic health records (EHRs) system organized by Taiwan's National Health Insurance Administration. This gives it a competitive edge in developing medical A.I. systems that use machine learning based on medical record data.
The Future of Healthcare: Global Collaborative Intelligence
The study was conducted in partnership with Harvard Medical School, Taipei Medical University, and Aesop, the first federated learning model for preventing medication errors that are optimized by combining models from multiple countries.
"Our AI model for medication safety has been trained by one of the world's largest prescription databases, 1.5 billion well-coded prescriptions from the U.S. and Taiwan, to learn the association between diagnosis, medication, and complex prescribing behavior of doctors from different countries. The study has shown the model trained by federated learning (FL) achieves remarkable performance comparable to the other two models trained by individual data sets," said Jim Long, CEO, and Co-founder of Aesop Technology.
Through implementing, the system can immediately provide adaptive suggestions to help the doctor better complete the prescription whenever physicians prescribe diagnoses or medications that cannot be explained. The new model has been deployed in several hospitals and has since been expanded to the eastern and western United States to catch medication errors before they make an impact.
"Data-driven medicine demands huge and diverse medical data sets. The biggest challenge is successfully implementing data-driven applications in clinical practice, from local to global, without compromising patient safety and privacy. FL provides the solution by training algorithms collaboratively without exchanging the data itself." said Dr. Yu-Chuan Jack Li, President-elected of the International Medical Informatics Association (IMIA) and Distinguished Professor of Taipei Medical University.
The result is a breakthrough in the international transferability of medical AI. It demonstrates a way to provide practical data-driven prescribing support to improve patient safety in the U.S., even though it could be challenging to obtain data to develop these systems locally.
Referred from: BENZINGA