ASGE at DDW AI in GI Workshop | May 2022-Debate, Discussion and Demonstration: Scientific Challenges of Contemporary Machine Learning as It Pertains to Gastroenterology and Solutions from a Technology Perspective

Debate, Discussion and Demonstration: Scientific Challenges of Contemporary Machine Learning as It Pertains to Gastroenterology and Solutions from a Technology Perspective

Back to course

Pdf Summary

In this document, Dr. Sravanthi Parasa discusses the future of artificial intelligence (AI) in endoscopy and gastroenterology. The main areas of focus for redefining healthcare include patient-oriented AI, clinician-centered AI, and administrative operations AI. Patient-centered care and interoperability are key to driving better outcomes with machine learning, enabling AI in healthcare through cloud and connected health systems, and improving access to care through digital transformation. AI can also help with physician fatigue by utilizing natural language processing (NLP), natural language understanding (NLU), AI scribes, and conversational AI.<br /><br />Deep learning algorithms have shown successful application in gastrointestinal endoscopy imaging, including lesion detection, lesion characterization, instrument detection, polyp size estimation, bowel preparation assessment, and adenoma detection rate (ADR) measurement. However, one of the challenges in machine learning is the insufficient quantity of training data. Data augmentation techniques using generative adversarial networks (GAN) can help create realistic artificial colon polyp images.<br /><br />Other applications of AI in healthcare include predicting falls, genomics, predicting length of stay, population health management, consumer health management, readmissions management, intelligent virtual care, precision medicine, smart IoT sepsis, drug development, drug prescription variation, staffing optimization, intelligent remote monitoring, inpatient deterioration, fraud, waste, and abuse detection, and no-show predictions.<br /><br />Remote monitoring using connected sensors can provide a more holistic view of a patient's experience by collecting data not only from traditional visits to healthcare facilities but also remotely from everyday life. This data can include both visible episodic data points and invisible continuous data points.<br /><br />As medical knowledge doubles in shorter time periods, AI can assist in building biomedical knowledge graphs and analyzing medical literature. AI can also play a role in proactive and data-driven clinical research, real-world data analysis, clinical trial design methodology, and understanding the gut microbial ecosystem in inflammatory bowel diseases.<br /><br />However, the impact of AI in clinical implementation has been limited due to a lack of implementation research, revenue generation, and focus on algorithm development rather than solving specific problems. Dr. Parasa encourages a paradigm shift towards leveraging machine learning to enable solutions to healthcare challenges. The document concludes with a quote from Albert Einstein emphasizing the importance of imagination in driving innovation.

Asset Subtitle

Sravanthi Parasa, MD and Scott Penberthy, PhD

Keywords

artificial intelligence

endoscopy

gastroenterology

patient-oriented AI

clinician-centered AI

administrative operations AI

patient-centered care

interoperability

machine learning

physician fatigue