A conversational artificial intelligence agent for medication reconciliation and review
Deo, R. C.; Goto, S.; Jain, T.; Meier, S.; Patel, R.
Show abstract
Medication reconciliation, the process of creating an accurate medication list for a patient, is critical to patient safety and care quality but requires clinical expertise and time. Large language models (LLMs), with the ability to generate language and respond to diverse prompts, oVer the potential to automate medication reconciliation and review, including via spoken conversation. We developed AMREC, the Atman Medication REconciliation Conversational AI agent. AMREC uses a fine-tuned version of Llama-3.1-8B-Instruct to standardize a patients medication list by extracting 18 elements from each prescription. A voice agent then follows two dialogue flows: 1) iterating through the medication list with identification and correction of discordances and 2) collecting and clarifying requisite details on any additional medications taken. The extraction model achieved an accuracy rate of 98.3% across prescription elements, and user testing demonstrated the conversational AI agents ability to confirm, correct, remove, and add new medications to a candidate list. With additional development, AMREC could be deployed in the context of frequent medication reconciliation, thereby improving patient care outcomes and reducing the high-cost burden of medication errors on the healthcare system.
Matching journals
The top 8 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Large Language Models in Real-World Clinical Workflows: A Systematic Review of Applications and Implementation 94%
- Listening to mental health crisis needs at scale: using Natural Language Processing to understand and evaluate a mental health crisis text messaging service 92%
- Development and Validation of a Machine Learning Model Integrated with the Clinical Workflow for Inpatient Discharge Date Prediction 92%
Similar papers in this journal
Similar papers in this journal
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.