×

News publications and other organizations are encouraged to reuse Direct Relief-published content for free under a Creative Commons License (Attribution-Non-Commercial-No Derivatives 4.0 International), given the republisher complies with the requirements identified below.

When republishing:

  • Include a byline with the reporter’s name and Direct Relief in the following format: "Author Name, Direct Relief." If attribution in that format is not possible, include the following language at the top of the story: "This story was originally published by Direct Relief."
  • If publishing online, please link to the original URL of the story.
  • Maintain any tagline at the bottom of the story.
  • With Direct Relief's permission, news publications can make changes such as localizing the content for a particular area, using a different headline, or shortening story text. To confirm edits are acceptable, please check with Direct Relief by clicking this link.
  • If new content is added to the original story — for example, a comment from a local official — a note with language to the effect of the following must be included: "Additional reporting by [reporter and organization]."
  • If republished stories are shared on social media, Direct Relief appreciates being tagged in the posts:
    • Twitter (@DirectRelief)
    • Facebook (@DirectRelief)
    • Instagram (@DirectRelief)

Republishing Images:

Unless stated otherwise, images shot by Direct Relief may be republished for non-commercial purposes with proper attribution, given the republisher complies with the requirements identified below.

  • Maintain correct caption information.
  • Credit the photographer and Direct Relief in the caption. For example: "First and Last Name / Direct Relief."
  • Do not digitally alter images.

Direct Relief often contracts with freelance photographers who usually, but not always, allow their work to be published by Direct Relief’s media partners. Contact Direct Relief for permission to use images in which Direct Relief is not credited in the caption by clicking here.

Other Requirements:

  • Do not state or imply that donations to any third-party organization support Direct Relief's work.
  • Republishers may not sell Direct Relief's content.
  • Direct Relief's work is prohibited from populating web pages designed to improve rankings on search engines or solely to gain revenue from network-based advertisements.
  • Advance permission is required to translate Direct Relief's stories into a language different from the original language of publication. To inquire, contact us here.
  • If Direct Relief requests a change to or removal of republished Direct Relief content from a site or on-air, the republisher must comply.

For any additional questions about republishing Direct Relief content, please email the team here.

After Disaster, AI Models Support Rapid Understanding of Health Facility Damage

News

Health

A man walks among the collapsed buildings in Maras, Turkey, days after severe earthquakes struck the country in February 2023. A new paper looks at the humanitarian use of AI and satellite imagery to quickly assess damage to health facilities. (Photo by Baran Ozdemir for Direct Relief)
By Olivia Lewis

Aerial satellite images paired with artificial intelligence can support humanitarian response during disasters, according to findings from a new paper published PLOS Digital Health.

Researchers compared how machine learning models can assess damage to health infrastructure from satellite imagery. The use of the models can help responders initially assess the damage within a geographic area to better coordinate post-disaster response. In a time and cost-saving measure, the models decrease the need for ground-based structural evaluations immediately following a disaster when life-saving measures are still being considered.

Researchers from Direct Relief and the Center for Innovation to Implementation published the article on PLOS Digital Health in October.

Researchers compared Microsoft and Google machine models to understand the structural damage caused by the 2023 earthquake in Turkey. They found that using AI models can help create rapid health facility damage reports. However, researchers also found that the models lack the ability to replace ground-level evaluations and underestimate damage to destroyed buildings long term.

Andrew Schroeder, Vice President of Research and Analysis at Direct Relief and one of the authors of the paper, said the models differ in how damage is defined. However, he finds the model’s potential promising, as it could identify population access to emergency medical supplies like pharmacies and doctor offices.

“It’s super impressive, but in need of more work,” he said. “In the future, we could aim to have a full-scale assessment of the change in the healthcare capacity of an area, remotely, within 42 to 78 hours after an event and without having to put assessors on the ground.”

“Predictive damage modeling has been around for several years now,” said Schroeder. “What’s new about the paper is the comparison of machine learning models for predictive damage assessment, specifically in the health sector.”

Filed Under: , , , , , ,

Giving is Good Medicine

You don't have to donate. That's why it's so extraordinary if you do.

Receive Alerts