Skip to main content

Artificial Intelligence and Machine Learning in Healthcare

Guest editors:
Theodoros Zanos : The Feinstein Institutes for Medical Research, USA
Douglas Barnaby : The Feinstein Institutes for Medical Research, USA
Marc Paradis: Northwell Health, USA
Andrew Seely : University of Ottawa, Canada
Shawn Stapleton : University of  Washington, USA

New Content Item (2)

Healthcare is experiencing a digital data explosion driven by widespread utilization of electronic medical records, development of innovative monitoring technologies, and increasing adoption of wearable consumer devices. As a result, interest in utilizing Artificial Intelligence (AI) and Machine Learning (ML) techniques to better understand and leverage these rich datasets to improve healthcare has grown exponentially. This presents a range of unique challenges as well as exciting opportunities specific to healthcare. In this thematic series, we will feature manuscripts that showcase the latest advances in this emerging field within the following broad areas of interest:

  • Dynamic, self-monitoring, and auto-updating clinical risk prediction models. Common approaches to risk prediction are often static and utilize cross-sectional, one-time capture of risk factors, rather than repeated longitudinal measurements that vary over time. They are also infrequently updated following their initial derivation and validation. A promising group of modeling frameworks has been developed to address these issues that include dynamic, self-monitoring, and continuously updated models designed for repeated longitudinal evaluations. A shift from static to dynamic risk prediction requires frameworks that incorporate temporal validation to detect performance drift, include retraining and recalibration schemes, and architectures that incorporate temporal trends and evolving risk factors.
  • Predictive modeling and the vision of the Learning Health System. Identifying relevant, real world clinical problems amenable to predictive solutions, defining optimal strategies to successfully implement these models, and monitoring all relevant outcomes post-implementation (e.g., model performance, clinician adoption, clinical effectiveness) are critical to the successful application of predictive modeling to advancing safe, efficient, and effective care. This is closely aligned with the paradigm of the Learning Health System, within which multidisciplinary teams (e.g., patients, researchers, clinicians, informaticists, biostatisticians, improvement & implementation scientists, industrial engineers) collaborate to continually learn, improve, and innovate as an integral part of the care process.
  • Validation frameworks for clinical predictive algorithms. Recently, several widely-adopted algorithms have been found to underperform when more broadly implemented and disseminated. Non-representativeness of training data, non-clinical optimization functions, site-specific implementation factors, weak change management approaches, and distributional drift often underlie failures to generalize. While validation frameworks exist for each of these issues, there is a need to establish a widely accepted framework that is consistent and integrated across the entire lifecycle of algorithm ideation, validation, and implementation. 
  • Predictive analytics and models based on continuous streaming data. Continuous high-frequency monitoring of clinical vital signs and waveforms is routine across emergency departments, cardiac telemetry floors and intensive care units, and is increasing being used on traditionally non-monitored wards. These data, while critical for real-time patient monitoring, are rarely retained and incorporated within predictive models, which represents a lost opportunity to incorporate the unique insights provided by these measures. A range of predictive analytics based upon such high-frequency data has shown promise in assisting clinicians with making critical decisions, and is a rapidly growing field that has garnered increasing interest in both academic and industry circles.

We welcome the submission of additional manuscripts to this series*.

*Articles must be submitted through Editorial Manager. Please indicate at the Additional Information stage of submission that you are submitting to the “Artificial Intelligence and Machine Learning in Healthcare” thematic series.

All manuscripts received will be subject to peer review as is standard for the journal and are subject to all of the journal’s standard policies. Articles will be added to the Collection as they are published.

  1. The fourth industrial revolution has led to the development and application of health monitoring sensors that are characterized by digitalization and intelligence. These sensors have extensive applications in ...

    Authors: Chan Wang, Tianyiyi He, Hong Zhou, Zixuan Zhang and Chengkuo Lee
    Citation: Bioelectronic Medicine 2023 9:17