Validation of Acute Pancreatitis Using Machine Learning and Multi-Site Adaptation for Anaphylaxis

A general methodological framework for developing improved health outcome of interest identification algorithms in the Sentinel Distributed Databases (SDD) using machine learning and natural language processing (NLP) techniques is being developed for the ongoing Sentinel project, “Validation of Anaphylaxis Using Machine Learning.”

This project will build on the anaphylaxis algorithm development work at one Data Partner and will have four aims:

• Aim 1: Extend and evaluate the applicability of the ongoing anaphylaxis algorithm development work at one Data Partner by scaling the algorithm and case identification procedures for use at a second Sentinel Data Partner. 
• Aim 2: Use and evaluate the ability of the general framework built for anaphylaxis to conduct chart review for acute pancreatitis.
• Aim 3: Use and evaluate the ability of the general framework built for anaphylaxis to conduct “deep annotation” of validated acute pancreatitis cases.
• Aim 4: Given the establishment of a “ground truth” of validated acute pancreatitis cases using expert medical chart review, use and evaluate the ability of the general framework built for anaphylaxis to conduct machine learning and utilize NLP techniques to develop risk prediction models. The aim is to improve the accuracy with which acute pancreatitis can be identified using structured and unstructured electronic data.