Development of an algorithm to automatically transcribe e-prescription directions in the pharmacy
The World Health Organization estimates that medication errors result in $42 billion in annual treatment costs. The lack of standardization in electronic prescription (e-prescription) directions across healthcare is a medication safety risk. For example, a recent national study identified 832 unique permutations for the direction concept “Take 1 tablet by mouth once daily”. Errors in the prescription directions can cause severe issues related to patient safety, such as adverse drug events, overdoses, or undertreatment of disease. There is a critical need to standardize the prescription directions agnostic to EHR systems, so as to decrease the miscommunication between prescribers and pharmacists, alleviate time pressures on pharmacy staff, and increase clarity of instructions to patients. The primary objective of our MCube proposal is to develop and validate a natural language processing algorithm that transcribes e-prescription directions received from prescribers into patient-friendly language. Accomplishment of this objective will lead to decreased prescription direction errors and reduce patient harm.
Presented at the Michigan Institute for Data Science Annual Symposium