International Organization for Migration (IOM)
International Organization for Migration (IOM)
Description of Activities on AI
On-going initiatives regarding data science methods (such as artificial intelligence (AI) or machine learning (ML)) within IOM’s Displacement Tracking Matrix (DTM) team, have two predominant work streams. The first focuses on developing ethics & guidance through inter-agency collaborations, and the second, as part of DTMs Global Internal Quality Control mechanisms for data management and analysis.
Project 1: Humanitarian Data Science and Ethics group (DSEG)
The group was established to coordinate and collaboratively identify the potential benefits and risks of advanced data science applications for the humanitarian sector and to establish and strengthen existing ethical frameworks and standards behind the use of these methods specific for humanitarian purposes. It encourages responsible data use.
Project 2: Using Collective Intelligence to Reduce analytical bias and introduce local participation in designing humanitarian response in drought-affected displacement contexts
The objective of this project is to utilize the unique insights that affected populations have into their situations and needs in order to pilot an improved aid delivery process. This experiment aims to demonstrate that the use of a collective intelligence approach in humanitarian data analysis can improve aid delivery by reducing the biases both in agency data processing and the participation of affected populations. We further seek to strengthen decision-making related to aid delivery by mitigating biases and process-flaws that result from sectoral or organizational mandates and expanding community representation, buy-in, and empowerment in the process. This experiment represents a proof-of-concept, exploring the impact of applying collective intelligence to DTM processes. If successful, this experiment will contribute towards the improvement of DTM’s data and analysis processes and prove the efficacy of CI in humanitarian response.
Project 3: Applying techniques for internal quality control within the Displacement Tracking Matrix (DTM) Global Team
DTM Global team Applies AI in Anomaly detection on migration data, and contextualisation of these data using #IDETECT, Rural / urban land classifications of displacement settings from DTMs central data warehouse, quality control routines (based on usual statistics, time-series models, NLP, aerial image recognition, etc.). AI is also applied in analysis of Drone imagery in displacement camps to facilitate a data-driven response to crisis severity measures on living condition in camps during natural disasters.
Project 4: The Big Data for Migration Alliance (BD4M)
The BD4M is the first-ever dedicated network of stakeholders seeking to facilitate responsible data innovation and collaboration to improve the evidence base on migration and human mobility and its use for policy making. Building relationships between governments, international organizations, and civil society to engage in migration policymaking will be key to effectively harnessing data innovation for migration policy. In order to accelerate the creation of new partnerships – data collaboratives – a set of guiding data responsibility principles must be agreed upon and implemented. The BD4M aims to actively address both the need to scale data collaboration and address the ethical challenges associated with using new data sources for migration.
Project 5: The Data Innovation Directory – a curated repository of innovative data applications on migration and human mobility
As part of the Big Data for Migration Alliance (BD4M), IOM’s Global Migration Data Analysis Centre (GMDAC)has developed a user-friendly, searchable curated repository of data innovation projects and initiatives in the area of migration and human mobility, including information about the project objectives, lead and partner organizations, focus topics, data sources, SDG or GCM objectives targeted, results to date, and links to further information, among others.
Project 6: Ten key policy questions related to migration, whose answers can be found in data and data science
Identification of ten key policy questions related to migration, whose answers can be found in data and data science. This is part of the wider “100 Questions Initiative” by the GovLab. The ten key questions on migration will be sourced by leveraging a community of “bilinguals” – practitioners across sectors globally who possess both migration and data expertise. The policy questions are meant to address the demand for data innovation.
Project 7: How Facebook Network data can contribute to identifying trends in migrant stocks in selected countries
An analysis of how Facebook data can contribute to identifying trends in migrant stocks in selected countries. It was triggered by the need to draw valuable insights from the vast amounts of data on human mobility resulting from exponential growth in the use of digital devices and internet services around the world.
Project 8: Analysis of Twitter data to measure anti-migrant sentiment during COVID-19
The proposed pilot project aims to measure and monitor changes in attitudes towards immigrants during the current COVID-19 outbreak using Twitter data and machine learning. Specifically this project seeks to a) Identify the key discrimination and racism acts and experiences undergone by immigrants in five countries: the United States, the United Kingdom, Spain, Italy and Germany; b) determine the extent of intensification in anti-immigration sentiment as the geographical spread and fatality rate of COVID-19 increases; and, c) assess how the key challenges, acts of discrimination and racism experienced by migrants vary by country.
Project 9: Analysis of Google Trends data to create forecasting tool for migration flows
The proposed pilot project aims to explore Google Trends data in order to develop a tool for policy makers to monitor in migration-relevant online searches and anticipate migration flows between countries and regions.
Project 10: Global Migration Data Analysis Centre – Strengthening national capacities to harness big data and novel methods for migration policy
IOM is proposing a programme to build national capacities in selected low- and middle-income countries to leverage new data sources, such as data from mobile phones, social media and satellite imagery, as well as new methods combining traditional and new sources, in migration analysis for policy.
Project 11: Research project collaboration to assess the Diagnostic accuracy of computer-aided detection solutions to identify pulmonary tuberculosis on Chest x-rays of TB screening cases
IOM and FIND entered a research collaboration in 2017 to conduct two parallel studies at the respective organizations to evaluate the accuracy of the AI solutions for the screening of suspected TB cases, after decision to have more information was made by WHO expertise meeting in 2016. The aim of the research project is to conduct AI evaluations independent of the developers using a similar study design and analysis plan, but with separate databases of CXR DICOM images with corresponding clinical and demographic data from individuals who underwent screening, and inform the result to WHO for guiding the WHO recommendation of using AIs for national TB Programs. The study at IOM assesses the accuracy of the AIs on HA of migrant bound to US, using sample of CXR DICOM images done during the health assessment and the retrospective clinical data. The study was conducted after getting approval from IOM Legal counsel and CDC, received ethical clearance from McGill University, and signed IOM legal agreements with FIND and all the three AI companies.
Project 12: Early and Improved Tuberculosis case detection and treatment among migrants and their families in provinces 1, 2 and 3 of Nepal through Public Private Mix Approach
IOM Nepal under the TB REACH project is planning to conduct an operational research to assess the impact of Artificial Intelligence on early and improved detection of Tuberculosis in Nepal. This is already discussed and agreed by the NTCC. For this operational research, the Qure.ai has provided in-kind support with 30,000 reading and 3 Qure.ai box (equivalent to $50,000) and will be linked with the three digital X-ray of National TB Center service sites where F.A.S.T. strategy is already introduced. Currently, the ITC global is reviewing the agreement with Qure.ai at technical level and we are hoping that it will be completed very soon. Meanwhile, we are also trying to have a meeting with the NTCC Director and his team to add COVID19 testing in the operational research as well. The findings of the study will be published in the peer-reviewed journal.