The aim of medication reconciliation is to ensure that healthcare patients receive a comprehensive and accurate list of their medications. Medication reconciliation efforts are largely manual and labor intensive processes, and measuring the effectiveness of those programs is equally labor intensive. Machine learning can be utilized to decrease the manual efforts and provide actionable data to understand where clinicians can focus their valuable time.

Pariveda is offering a fixed fee proof of concept to healthcare customers looking a qualitative snapshot into the effectiveness of their Medication Reconciliation programs. As part of the proof of concept, the customer will receive:

• 8 machine learning models for predicting common medication reconciliation errors
• A report with a snapshot of the current error rate with improvement recommendations

Customer will be required to complete the following activities prior to the commencement of the proof of concept

• Extract and de-identify data for 100K medications from their Electronic Health Record system
• Curate 30K medications for medication errors to create a ground truth dataset for model training

AWS products used include:

• Amazon SageMaker
• Amazon S3
• Comprehend Medical
• Amazon QuickSight (optional)
• IAM (user accounts/permissions)