Last month’s earthquakes in Turkey’s southern Kahramanmaraş province were among the most deadly of the century. Since they first struck, the quakes have resulted in the deaths of over 51,000 people in both Turkey and Syria. UN OCHA estimates that they caused around 120,000 injuries, which may be a drastic undercount.
An estimated 3 million people have been displaced throughout the region. AFAD, a governmental disaster management agency operating under the Turkish Ministry of Interior, reported that as of February 25th 1.6 million people were being sheltered in the most-affected areas. Moreover, 323,000 were being sheltered outside of the affected areas and around 900,000 were sheltering elsewhere under their own means.
The International Federation of Red Cross and Red Crescent Societies (IFRC) reported that 2.2 million building units have sustained damage. Given the average occupancy in the region is of three people per household, the agency estimates that between seven-to-eight million people lived in homes that were at least partially damaged by earthquakes.
In the case of Syria, the disaster comes at a time when the country is already experiencing a longterm humanitarian crisis. Already displaced populations, the underlying weakness of the country’s health system, and a preexisting cholera epidemic all pose a severe long-term set of risks for those affected by the disaster.
On March 3, 2023, CrisisReady’s “Data in Crises” event, co-hosted by the Harvard Data Science Initiative (HDSI), convened a group of experts working in response to the earthquakes to discuss their efforts.
Speakers from the Humanitarian OpenStreetMap Team (HOT), the Syrian American Medical Society (SAMS), Yale University, and NeedsMap Social Cooperative joined the session alongside CrisisReady’s co-directors Dr. Caroline Buckee, who’s also a professor of epidemiology at Harvard T.H. Chan School of Public Health, and Andrew Schroeder, who’s also the vice president of research and analysis at Direct Relief.
Together they explored ongoing relief operations, focusing on how data is being used in response.
Novel data and advanced analytic methods have played an important role in the response to the earthquakes, particularly for:
Assessing earthquake damage
Satellite imagery and remote sensing data were used to assess the extent of earthquake damage in affected areas. This information was then used to identify areas where aid was most needed.
One of the most notable organizations using novel in damage assessment Humanitarian OpenStreetMap Team (HOT), which has been creating detailed maps of the most-affected areas in Turkey to assess the extent of the damage.
HOT has also been working closely with local communities and aid organizations to identify priority areas for relief efforts and allocate resources more effectively. In addition, they have been using machine learning algorithms to automate the detection of damaged buildings and infrastructure, which has significantly accelerated the assessment process.
Monitoring population displacement
Data from mobile phones, social media, and other sources were used to monitor population movement and displacement patterns. This information was used to help aid organizations and government agencies to better understand where and when to allocate resources more effectively.
Working in this area, CrisisReady research team has been gathering and analyzing data on population displacement and mobility patterns. Situation reports published by the team, which provide near real-time insights into population movement and identify areas where displaced individuals are congregating, have helped aid organizations target their relief efforts and ensure that aid is delivered to those who need it most.
Predicting future needs
Data analytics and machine learning were used to develop predictive models that could help anticipate future needs and inform longer-term disaster recovery efforts.
Coordinating relief efforts
Data sharing and coordination platforms, such as NeedsMap Social Cooperative, have played a pivotal role in facilitating the collaboration and sharing of information among aid organizations, local responders, and other key stakeholders. The company is also mobilizing support for displaced individuals through an online platform that allows locals to post vacant housing units or rooms available to rent.
Tending to the health care impacts of the earthquakes, Direct Relief, a humanitarian aid organization that provides medical supplies and resources to communities affected by disasters, and the Syrian American Medical Society have been working to deliver critical medical supplies and resources to affected communities. This includes distributing medications, medical equipment, and other essential supplies.
The two organizations have also been working to provide support to local health systems and medical facilities that have been overwhelmed by the influx of patients following the earthquakes by providing funding for medical personnel, equipment, and other resources to help these facilities provide care to those in need.
Overall, data have been crucial in response to the earthquakes in Turkey and Syria, yet several challenges remain. Given the acute challenges of conflict and displacement in the region, all humanitarian data applications must grapple with serious legal, ethical, and technical challenges.
The following excerpts below are taken from conversations that took place during the “Data in Crises” event on March 3, 2023.
The Ethical Challenges of Using Data in Humanitarian Contexts
The use of personally identifiable information (PII) in humanitarian response can raise several ethical concerns, among which are issues of privacy and security, consent, transparency in the use of such information, and fairness in its representativeness. The use of this information also comes with the potential for harm, exposing groups of individuals to exploitation, exclusion, and mistreatment.
The ethical responsibilities of data scientists who gather, analyze, and synthesize PII into actionable products, as well as the responders and officials who use them to inform operations and policy, are vast and highlight the need to identify the evolving risks associated with temporal, spatial, and demographic data throughout the data preparedness cycle.
How data scientists adjust the legibility of such data is typically the largest control they have in mitigating — and perhaps preventing — ethical risks by engaging in data minimization. In this context, data minimization relates to the limits placed on the collection and retention of personal information to what is directly relevant and necessary to accomplish a specified goal.
Nathaniel Raymond, who led discussion on the ethical challenges of using sensitive data in response contexts, stresses the importance of data preparedness planning in ethical triage prior to response interventions. He says that when a robust data preparedness plan is developed and adhered to, responders are able to minimize data across the the aforementioned “buckets” of data; those relating to time (temporal), place (spatial), and people (demographic).
As you build a data preparedness plan to have ethics into it, you’re really having to demount — to a context, to a scenario — how your temporal loops change across a disaster. So, what it was in the first 24-hours is not the same as it is now.
And your risks change with temporality in terms of that resolution and legibility. Same with spatial, and same with demographic.
How you turn the knobs to make more legible and less legible the time, place, and people elements […] is usually the largest control you have on either mitigating ethical risks or often […] preventing ethical risks by engaging in data minimization. Often the largest tool you have is not collecting in the first place.
More Quotes from Nathaniel Raymond
There is no disaster, whether it is “natural” or armed conflict, that is not a protection challenge in terms of how data can affect the vulnerability of populations at risk.
There is sometimes an approach which treats data in natural disasters as somehow with less risk than data in armed conflict.
As we can see from this example with the protection issues and political issues in Turkey, and also the armed conflict issues in Syria, we need to treat this data as if we were in an armed conflict. Often, that’s the case in almost every single disaster.
The ethical responsibilities and ethical challenges are constant, regardless of whether it’s an armed conflict or not.
Your ethical responsibilities of using this data sit at every stage of the data lifecycle. Ethics is a process, it’s not a product.Nathaniel Raymond
PII — personally identifiable information — is where most of our ethics in academia focus. In disasters PII is important, but often it is not the vector for harm or the pathway to harm in the way that DII — demographically identifiable information — about sub-demographic groups within a cohort that’s been affected by an emergency.
Often those that are looking to exploit, to target, or to exclude — those are the three types of big harm that happens in data — they’re not looking for individuals, they’re looking for demographic groups. And so, as part of your planning across this lifecycle — and as part of your data preparedness plan— you need to be identifying what populations do you make legible.Nathaniel Raymond
Data preparedness planning so crucial to being ethical in how you [respond to a disaster] because often we are pressed in a disaster to be maximalists, to collect everything. When we do data preparedness planning in ethical triage — in ethical audit before we go in — we end up being able to data minimize more effectively.
It’s that minimization across those three baskets — time, place, and people — that actually protects populations, whether we’re talking an earthquake, whether we’re talking a hurricane, or a war.Nathaniel Raymond
Using Data for Infectious Disease Modeling and Forecasting During Disasters
Epidemiological modeling during disasters is critical to understand the overlapping mechanisms and pathways by which disasters and armed conflicts lead to infectious disease outbreaks. The need for modeling in these contexts is especially important given the elevated risk of transmission by common disaster hazards, such as population displacement, infrastructural damage, overcrowding, and poor access to clean water.
In the case of the earthquakes in Turkey and Syria, infrastructural damages to local health systems have been especially harmful. With these damages comes a set of short-term trauma impacts and longer-term disease and mortality impacts. Both sets of impacts disproportionately affect vulnerable populations, such as those with electricity-dependent medical needs, chronic diseases, and elderly people who are more susceptible to infection.
Synthesizing data into deliverable information products to address the risks of infectious disease outbreaks in disaster-affected populations can be incredibly difficult, explained Professor Caroline Buckee, who led discussion on epidemiological modeling in humanitarian relief contexts at the Data in Crises event. This is especially true given how quickly these models need to be generated in the midst of a response.
Addressing the core challenges of infectious disease modeling during disasters involves improving the translation of analytic processes into actionable insights for field response. This translation process is summarized in the feedback loop below, which Professor Buckee shared during her presentation.
The first component of the system involves defining populations at risk on relevant spatial scales. She explained that many of the infectious disease models and forecasts produced in the wake of these events replicate data on population density. Data on population densities typically come from census data, which can be outdated and unreliable in and of itself, and even when newer data sources are used, displacement can make it difficult to discern accurate baseline population figures used to quantify the risk of disease outbreaks in a given area. This represents a major analytic challenge, not only in regards to the development of forecasting models, but in understanding what segments of a population are most at risk and where resources need to be allocated in the immediate term.
The second component of the feedback loop deals with determining how many cases of a particular infection there are at a specific time and place. This gives rise to another challenge in infectious disease modeling. Without knowing the accurate disease count, epidemiologists cannot produce reliable disease forecasts using mechanistic approaches, she said, and statistical modeling essentially becomes guesswork when diagnostic and reporting systems are unavailable. In order to have reliable disease forecast models, Professor Buckee stresses the importance of establishing data access processes and partnerships before the disaster takes place so that data flows and information sharing methods are well-formed.
Without knowing what the count of disease cases are you’re not going to be able to make a good disease forecast using a mechanistic approach. […]
It’s often the case [in] forecasting the single biggest challenge because without the availability of diagnostics and reporting systems you’re basically guessing.
And so, in order to reasonably have things like disease forecast models you actually need data access and partnerships before the disaster, so that those data flows have been figured out […]
The third component of the process — running the forecasts — typically doesn’t pose many risks as long as the previous steps are fully fleshed out, stated Buckee. However, issues of validation can be a problem during this phase.
The last piece of the framework, which entails communicating risk maps effectively to decision makers, tends to be the weak link in the process. Professor Buckee explained that this is largely because epidemiologists may not have the speed to communicate risk in an iterative way. Without continuously updating data on disease prevalence, forecasting efforts fall short. As is true in the second phase of the framework, the development of partnerships is essential to ensure that disease models are of service to key decision makers. “You don’t just make one model forecast or make a dashboard and give it to people,” said Buckee, “you have to go back and provide updated information — talk to them about what they need — and that takes time and needs to be done in advance sometimes.”
More Quotes from Caroline Buckee
…My three takeaways for translation of analytic and data products in the context of crises are first, that preexisting partnerships have to be there because you can’t build them at the same time that everybody is rushing around trying to respond to the crisis. […]
It’s often the case, in my opinion, that simple models are just much more flexible, more easily interpreted, more robust than very complicated approaches to trying to make forecasts or do clever analytics. […]
And the last is obviously that the decision-making needs on the ground have to drive the analytics. And that’s something that still is not commonly done, especially in academic circles. So I think with some of these translational pipelines and partnerships, we really have to make sure that we are tailoring the analytics an effective way.Caroline Buckee
To make disease modeling relevant for decision making we still need to really work on partnerships.
And I put an arrow going back to “defining the populations at risk” because part of that is an iterative process. So, you don’t just make one model forecast or make a dashboard and give it to people […] you have to go back and provide updated information, talk to them about what they need. […] That takes time and it needs to be done in advance sometimes.Caroline Buckee
At the moment, there’s often a disconnect between where the analytics and science in data are and what decision makers on the ground actually need […]
So defining the populations at risk is challenging in the context of displacement because we want to know as much as we can about demographics […]Caroline Buckee
For infectious diseases a lot of the models and forecasts we can produce replicate population density, and so the data we have on population density is often from census data — it’s often outdated. We can use new data sources, but when displacement happens, we don’t necessarily know what our denominators are.
That becomes one of the analytic challenges, not only with calculating what to expect but also with understanding who needs the prevention and where the resources need to go in the immediate term.Caroline Buckee
Complexities of Medical Response in Syria: Navigating Humanitarian, Public Health, and Economic Challenges
Dr. Abdulfatah Elshaar, an internal medicine physician in Massachusetts and chairman of the Syrian American Medical Society (SAMS), joined the Data in Crises event to share how the organization contributed to medical responses after the disaster. Also participating in this conversation was Dr. Samer Attar, who volunteered with SAMS in early February to provide clinical support to destroyed medical facilities in northern Syria.
Dr. Elshaar explained that the earthquakes represent a hybrid disaster for the country, compounding the impacts of preexisting humanitarian, public health, and economic crises. The ongoing Syrian civil war, which began 12 years ago, has significantly hampered essential health services due to damages sustained by medical facilities. More recently, the 2022 cholera outbreak and the lingering impacts of the COVID-19 pandemic have added further stress to the nation’s weak health system.
The displacement of healthcare professionals in Turkey and Syria exacerbated the situation. As of early March 2023, Dr. Elshaar stated that around 40% of northern Syria’s medical staff were displaced, highlighting the need for critical medical support after the disaster.
Immediately following the earthquakes, SAMS’ headquarters in Washington D.C. established a multi-pillar task force to coordinate relief efforts with the organization’s offices in Syria. The task force oversaw all response programs and operations, information management processes, the mobilization of essential resources, fundraising efforts, media advocacy, and the deployment of medical personnel from the United States.
Forming partnerships between SAMS and Syrian NGOs has been essential in addressing the non-medical needs in the area. SAMS continues to work closely with The White Helmets, a volunteer organization that operates in parts of opposition-controlled Syria, Direct Relief, and the Syrian Forum, a nonprofit organization that provides relief to displaced communities.
In term[s] of the continuation on the response that SAMS took, SAM established a partnership with other NGOs in the region […] to help with the non-medical need[s] …
They did an alliance with White Helmet, Direct Relief, and Syrian Forum to help address the immediate need given the magnitude of the crisis …
Dr. Abdulfatah Elshaar
Within the context of medical response, Dr. Elshaar explained that the earthquakes exposed significant gaps in organizational communication systems. It is crucial to address the challenges of exchanging critical health information between organizations, people, and technologies to develop effective programmatic interventions. This includes data on mortality, injuries, and disease transmission, as well as data to forecast the stock consumption of medical supplies and medications. Such information will help inform timely procurement decisions as the burden on facilities becomes more severe.
Dr. Elshaar emphasized the importance of defining a clear strategy for information gathering and sharing during a crisis. While many humanitarian organizations have strategies in place, they often lack the capacity to handle the urgent need for updated information in emergencies. Leveraging collaborations between partnering response teams, with their diverse sets of skills and expertise, could be especially helpful in supporting the rapid analysis and visualization of key data.
Healthcare Challenges in Post-Earthquake Syria: Addressing the Impact of Preexisting Crises and Urgent Medical Support
Following Dr. Abdulfatah Elshaar’s explanation of the healthcare infrastructure in Syria and the compounding crises the country faced prior to the earthquakes, Dr. Samer Attar shared his experiences in providing medical support in northern Syria after the earthquakes. Dr. Attar volunteered with SAMS in early February, joining a group of nine doctors, including surgeons, anesthesia doctors, pulmonologists, and nephrologists. They all packed up and met in Chicago to embark on their mission. However, their supplies were limited, and each of them carried only two to four bags. Upon crossing the border into Syria, they were immediately confronted with the grim reality of the disaster as they saw bodies of Syrians who had died in Turkey being repatriated back to Syria.
One of the hardest-hit areas was the Village of Genetas in northern Syria, which they visited and witnessed the immense destruction. Dr. Attar described the desperate attempts of people digging through the rubble with their bare hands to find their loved ones. The medical facilities in idlib City, where they mainly worked, were overwhelmed with patients suffering from crush injuries, requiring urgent care, fasciotomies, and dialysis.
The challenges they faced were immense, and Dr. Attar recounted the challenging triage decisions they had to make with limited resources. He expressed the difficult experiences of not being able to attend to everyone’s urgent medical needs in a timely manner. Among the patients were children and a young girl diagnosed with bone cancer but unable to access the necessary treatment.
Dr. Attar emphasized the urgent need for wound care, amputations, and prosthetics as these were already issues before the earthquake struck. He also pointed out the importance of addressing non-earthquake medical needs such as diabetes, hypertension, cancer, maternal fetal health, and cholera, which have become even more challenging to manage due to the disaster’s impact on the already weakened health system.
Beyond the physical injuries, Dr. Attar highlights the often-overlooked issue of mental health, affecting not only patients but also the nurses, doctors, and rescue workers on the ground. The trauma they experience while providing care in crisis settings is immense, and support for their mental well-being is essential.
Dr. Attar expressed his admiration for the strength and resilience of the people he met in Syria and Turkey. He acknowledged that in settings like these, logisticians, data analysts, supply chains, and resource management play a critical role in saving lives, often more than doctors. While the challenges are vast, he recognizes the ongoing work that needs to be done to address the immediate and long-term medical needs in the aftermath of the earthquakes in Syria.
Collaborative Efforts and Effective use of Open Mapping Data to Respond to the Earthquakes in Turkey and Syria
Can Unen, an OpenStreetMap trainer and active member of open mapping communities, shared insights into the collaborative response to the earthquake disaster in Turkey and Syria. The mapping activation began after the earthquakes struck, and they collaborated with various organizations to gather ground data in the affected regions. The efforts resulted in significant contributions, with over 1.6 million buildings and 70,000 kilometers of roads mapped, thanks to the involvement of more than 7,600 individuals. This massive data collection became an invaluable resource for agencies and humanitarian workers in the region.
Can emphasized the importance of data coordination among different actors, both national and international, to ensure efficient recovery efforts. He mentioned the abundance of data and actors in the field, but also the need for better coordination among them. As a part of the open mapping community, their priority was to bring together these different entities, ensuring they work collectively and efficiently. Can explained that by coordinating these efforts, they could address the gaps and challenges in the existing data.
He highlighted the significance of accurate and up-to-date data for recovery efforts. He stressed the need for the OpenStreetMap community to ensure their data reflected the ground situation as accurately as possible. In doing so, they could support agencies, NGOs, and displaced people in the affected areas. Can also mentioned the use of open data sources like Open Aerial Map, which provided satellite imagery that was crucial for the identification and assessment of collapsed buildings and damage.
Connecting Individuals with Organizations During Times of Crisis
In her presentation, Özge Acar introduced “NeedsMap,” an online platform that connects individuals and organizations willing to support those in need through online forums and maps. The platform’s main mission is to foster cooperative solidarity in various areas, including education and disaster response. Özge highlighted the significance of using technology and maps to bridge the gap between supporters and those requiring assistance.
During the earthquake disaster response, NeedsMap initiated mapping tasks to assess damages in Turkey and Syria. They employed a damage assessment form shared through social media to gather real-time information from the affected areas, where verified crowd-sourced data was used to ensure data accuracy and reliability. The organization then produced the “Disaster Response Map,” which provided four components, including the earthquake map, estimated damage buildings map, volunteer locations, and the assessment of damaged buildings.
Additionally, NeedsMap developed another product called “One House, One Rent.” This website allowed people to either request support or provide rental assistance, eliminating the need for transaction companies and ensuring direct support for those in need. Over 413 million Turkish liras were collected, and 3038 houses were shared, providing housing assistance to approximately 35,000 people affected by the earthquake.
However, Özge also mentioned the challenges faced during the disaster response efforts. Confirming needs and maintaining a balance between the materials in the warehouse and distribution are crucial yet difficult tasks. Volunteers’ mental health and well-being were essential considerations, as the emotional toll of assisting disaster victims often led to fatigue and depression. Communication through social media and WhatsApp brought forth additional challenges, as confirming needs and coordinating assistance required constant updates and cross-checking.
Despite these obstacles, NeedsMap remained committed to its mission, striving to bridge the gap between those in need and supporters through efficient technology-driven mapping and coordination efforts.
Question and Answer Segment
Below are a few questions asked by audience members of the Data in Crises event, as well as answers provided by the speakers they were given to.
Question: What sort of real-time information is critical for the Syrian American Medical Society’s staff to have after a crisis, and what are the sources and conditions surrounding real-time data?
Answer: After a crisis, SAM’s staff needs real-time information on the extent of damage, the location of affected areas, and the immediate needs of the affected population. This includes data on the number of collapsed buildings, medical facilities, and settlement information. The sources of real-time data can include social media posts, damage assessment forms, and data from institutions like universities. Access to this information is crucial for effective decision making and resource allocation during the response.
Question: How can we ensure that medical facilities continue to address the needs of those in crisis, and do we have the necessary data to redirect patients and ambulances efficiently?
Answer: To ensure that medical facilities can continue addressing the urgent medical needs, access to real-time data is vital, explained Dr. Elshaar. This data should include information on the location and capacity of medical facilities, the availability of medical supplies, and the current health needs of the affected population. With this data, medical staff can efficiently redirect patients and ambulances to the most appropriate facilities, ensuring a coordinated and effective response.
Question: What are the current conditions for modeling cholera post-disaster, and what data do we have and need for this process?
Answer: Dr. Caroline Buckee explained that the analytics for modeling cholera post-disaster are available, but the challenge lies in the translational loop and data sharing agreements. The necessary data for modeling cholera includes baseline maps, information on pre-existing risks, and age distributions of the affected population. To improve cholera modeling, Professor Buckee said that there needs to be well-defined frameworks, data sharing agreements, and an understanding of specific intervention needs for each situation.
Question: How should data in terms of reliability and ethics be approached, given the abundance of data and reports produced by agencies after the earthquake in Turkey?
Answer: Can Unen explained that the abundance of data after a crisis can lead to data repetition and inefficiency. It is crucial to coordinate data efforts and ensure a well-structured approach to data sharing and openness. While geographical and physical data may be abundant, he said that there should be careful verification of sensitive and personal information. Establishing clear data sharing protocols and coordination mechanisms can avoid duplication of efforts and ensure the reliability and ethical use of data.
Question: What are the challenges in accessing critical data during a crisis, and how can data flow be better coordinated and structured to enhance response efforts?
Answer: Can Unen said that the challenges in accessing critical data during a crisis include a lack of well-structured data sharing mechanisms and confusion over where to find relevant data. Coordinating data flow and openness is essential to prevent repetitive efforts and ensure valuable time is used efficiently. Again, Mr. Unen explained that establishing data-sharing agreements and clear pathways for information exchange can enhance response efforts and support decision making during a crisis.
Data analytics and novel data sources have proven to be an effective tool in responding to natural disasters around the world. For example, in the aftermath of earthquakes in Turkey and Syria, satellite imagery and other remote sensing data were used to identify areas with the most severe damage. In addition to this, mobile phone data has been utilized to track population movements and predict the spread of infectious diseases in disaster-affected areas. As proven by the discussions held at our “Data in Crises” event, these types of data can support disaster response, helping organizations synthesize accurate and up-to-date information resources and allocate resources more effectively. However, several challenges remain.
The majority of these challenges lie in the translation of complex data products into field response. This is especially true when it comes to epidemiological modeling, identifying resource needs, conducting damage assessments, and responding to population medical needs. While novel data provide valuable insights, it can be difficult to translate this information into actionable response efforts. In addition to these operational challenges, the use of sensitive data in humanitarian contexts raises several ethical and legal considerations. The use of personally identifiable (PII) data and demographically identifiable data (DII) raise concerns about privacy, confidentiality, and security.
Despite these challenges, the potential benefits of using novel data in response are immense. In order to fully realize their utility in these contexts, it is important to continue to address the ethical and legal considerations associated with the use of sensitive data in humanitarian contexts.
The “Data in Crises” event recording features all of the conversations held at the event, which delve into these issues and more. This is just one example of the valuable discussions taking place in the field of disaster response. Stay tuned for more “Data in Crises” events to come!