From self-driving cars to smartphones, artificial intelligence (AI) has already changed many aspects of human life. Some areas of health care, such as the analysis of medical images, are already benefiting from AI. But the role of AI in the health sector is still not clear. The visualizations accessible from this page have been designed to obtain a better understanding of the applications of AI in health care and of the available evidence about their benefits. The visualizations can be used to explore two related sets of questions:
What types of applications of AI to health care are already available or are expected to be available in the next five years?
We provide a view of the health conditions they target and the objective they want to achieve. In addition, we show the intended main users and whether the applications have or need an FDA clearance. This visualization is accessed by clicking on the tab named AI Application below.
Each symbol in the panel corresponds to an AI application. The health conditions targeted by an application are listed vertically on the left side of the panel. In the view displayed the first time the visualization is used, the purpose, or function, served by the application is listed horizontally at the top of the panel, under the heading Function.
We use shapes to represent users, such as patient, caregiver or health professional. We use colors to represent current development status (e.g., is the application still in development, does it have an FDA clearance, or is it already in use without need of an FDA clearance?). Moving the mouse over a specific symbol will open a tooltip that displays additional information about the application, including its name, the name of the developer, and a link to the corresponding website.
It is possible to change the information displayed across the columns of the grid from the default setting of Function by clicking on the tabs at the bottom. The Status tab will display the development status of the application, while the tab named AI Category will display the type of AI algorithm used by the application. Clicking on the heading of a specific row or column will highlight that row or column.
What type of evidence is available on the benefits and the accuracy of these application?
This evidence map shows, for a number of health conditions, the publication type of the evidence found in the literature, along with the study design and sample size used to obtain that evidence.
This visualization is accessed by clicking on the tab named The Evidence Base below. Each symbol in the panel corresponds to a study containing evidence regarding an AI application in health care. The health conditions targeted by an application are listed vertically on the left side of the panel. In the view displayed the first time the visualization is used, the type of publication is listed horizontally at the top of the panel, under the heading Publication Type. Each publication type is displayed with a different shape, and colors are used to represent the sample size of the study. Moving the mouse over a specific symbol will open a tooltip that displays additional information about the evidence, including the citation, a short summary with the author conclusions, and a link to the corresponding website.
Clicking the Study Design tab at the bottom will switch to a similar view in which we display the design of the study rather than the publication type. Clicking on the heading of a specific row or column will highlight that row or column.
Methodology (Summary Report)
Girosi F, Mann S, Kareddy V. Narrative Review and Evidence Mapping: Artificial Intelligence in Clinical Care. Patient-Centered Outcomes Research Institute; February 2021. Prepared by ECRI under Contract No. IDIQ-TO#12-ECRI-SCI-EVIDENCEMAP-2019-07-15.
View more details about the project here.
Artificial Intelligence: New Report Explores Applications for Health Care
Artificial intelligence is impacting the healthcare field, but its role outside medical imaging remains unclear. A new report and evidence map examine the evidence on non-imaging artificial intelligence applications in health care.
To share your feedback on this evidence visualization, contact the Research Synthesis team.