ASGE Masterclass: Artificial Intelligence (Live Virtual) | June 2021-09-taposh

09-taposh_ai-ehr-asge-2021-v3 (1)

Pdf Summary

In this document, Taposh Dutta Roy, an expert from Kaiser Permanente, discusses the use of artificial intelligence (AI) in predictive modeling of electronic health record (EHR) data. The goal of AI in EHR is to provide insights at both individual and population levels. The AI models require diverse types of data, including structured text, unstructured text, image data, video data, and audio data.<br /><br />The complexity of healthcare data is discussed, and various measures are suggested to define complexity in healthcare systems. The author emphasizes the importance of using degrees of interrelatedness of system components to define complexity.<br /><br />Different machine learning models for EHR are explained, including classification, regression, clustering, and reinforcement learning. The author provides examples of how these models can be used in healthcare applications, such as predicting the presence of cancer based on medical images or predicting the risk of early recurrence after surgery.<br /><br />Opportunities for AI in healthcare are discussed, including the need for organized, unbiased, transparent, explainable, and trustable models. The author also presents a concept for an AI solution in EHR data, involving the creation of cohorts and the use of unsupervised machine learning techniques.<br /><br />The document concludes with references to academic studies and papers that support the use of AI in healthcare, including the application of AI in gastroenterology and the prediction of cervical cancer using electronic health records.<br /><br />Overall, this document provides an overview of the use of AI in predictive modeling of EHR data, highlighting its potential for improving healthcare outcomes and decision-making.

Keywords

artificial intelligence

predictive modeling

electronic health record

EHR data

machine learning models

healthcare applications

cancer prediction

unsupervised machine learning

gastroenterology

cervical cancer prediction