Designing nanotheranostics with machine learning

Abstract

The inherent limits of traditional diagnoses and therapies have driven the development and application of emerging nanotechnologies for more effective and safer management of diseases, herein referred to as ‘nanotheranostics’. Although many important technological successes have been achieved in this field, widespread adoption of nanotheranostics as a new paradigm is hindered by specific obstacles, including time-consuming synthesis of nanoparticles, incomplete understanding of nano–bio interactions, and challenges regarding chemistry, manufacturing and the controls required for clinical translation and commercialization. As a key branch of artificial intelligence, machine learning (ML) provides a set of tools capable of performing time-consuming and result-perception tasks, thus offering unique opportunities for nanotheranostics. This Review summarizes the progress and challenges in this emerging field of ML-aided nanotheranostics, and discusses the opportunities in developing next-generation nanotheranostics with reliable datasets and advanced ML models to offer better clinical benefits to patients.