Advancing Point-of-Care Testing by Application of Machine Learning Techniques and Artificial Intelligence

Abstract

The promise of artificial intelligence (AI) has generated enthusiasm among patients, healthcare professionals, and technology developers who seek to leverage its potential to enhance the diagnosis and management of an increasing number of chronic and acute conditions. Point-of-care testing (POCT) increases access to care because it enables care outside of traditional medical settings. Collaboration among developers, clinicians, and end users is an effective best practice for solving clinical problems. A common set of clearly defined terms that are easily understood by research teams is a valuable tool that fosters these collaborations.We present brief, accurate, and clear descriptions of terms and techniques used to develop new device and decision support technologies in association with their most common applications to POCT. This lexicon of terms used to describe AI and machine learning techniques is quick reference for healthcare professionals, researchers, developers, and patients. Commonly used methods and techniques are tabulated and presented with text providing the context of their common usage and required data characteristics. Finally, we summarize model effectiveness measurement and the assessment of component features contributions.Artificial intelligence (AI) refers to non-human techniques that infer meaning from sets of data. It can produce generalizations, classifications, predictions, and can identify associations using automated learning methods. This guide provides an overview of these methods and their application to point-of-care testing.