Authors: Regine Mugeni, Aurore Nishimwe, Viviane Akili, Marc Twagirumukiza, Charles Ruranga
Tuesday, 28 March 2023
This blog is a reflection on the LAISDAR project's highlights, lessons learned and challenges that emerged for data harmonization, sharing, accessing and analyzing. The sources of data used are a country wide longitudinal survey in communities, an essay of data linkages from 15 hospitals/health centers and COVID-19 testing centers.
The COVID-19 virus has resulted in high morbidity and mortality rates all over the world. The preventive measures and management were strengthened by government authorities in Rwanda, however, the number of new cases continued to increase at that time of the outbreak and the screening was limited to individuals who came into contact with persons with COVID. As the pandemic progressed, there was a spike in the need for data to better understand the epidemiological profile of COVID in the region, but with the data recorded across diverse institutions, ranging from hospitals to the established COVID testing centers, it was a challenge to analyze the data. The main solution identified was to create a harmonized data system that could inform public health policies for predicting future cases and possible new infectious diseases.
The first part of the project focused on progressing data harmonization, for which the main approach was to integrate different data systems from various sources of COVID data (both longitudinal survey data and electronic health records). The process started with workshops with partners and stakeholders to explain the new technique (Artificial Intelligence) to be used in order to facilitate a smooth data linkage. Following this, the next step involved training information technology (IT) staff from the hospitals about the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). OMOP is an open community data standard to be used. When asked about the experience, Charles Ruranga, Principal Investigator of the LAISDAR Project said, “it was challenging to harmonize data across different sources mainly due to datasets stored in different formats but with support from experts in the consortium this has been solved and the project has been implemented smoothly.''
Good progress was made with the distribution of the tool, demo Mac Mini machines, that will be used for data harmonization among the hospitals. So far, the two demo Mac Mini machines and all Mac Minis were deployed to 13 hospitals and have been configured to run the necessary processes for hosting a common data model with synthetically generated COVID-19 data. However, being used for the first time to get the needed update and finalize the configuration took more time and affected the progress. The training of the IT team and hospital leaders on the Mac mini manual and instructions, as well as follow up practices have been organized on sites to overcome this challenge.
Training IT officers on 27 January 2023
The second part of the project involved a longitudinal survey for which data was collected over a period of 12 weeks; however, due to a series of lockdowns in Rwanda, there were extensive delays in starting data collection through community surveys. For the whole period of data collection 26,412 phone calls were made by data collectors. Sometimes they were unable to reach participants due to limited network connection.
“At the beginning of the project, Rwanda had a series of lockdowns due to COVID-19 and office work was stopped in various institutions including the University of Rwanda. Consequently, this delayed the smooth running of project activities, mainly the data collection aspect including getting authorization to use datasets from concerned hospitals, installation of video conferences, and initiating hospital data harmonization. As of today, the project is progressing well and the great achievements can speak for themselves (publication of papers, completion of data collection, hospital data harmonization, dissemination of the project…)” Viviane, AKILI Operations Coordinator said.
After the data collection period, a team of statisticians proceeded to do the data cleaning, which was completed in consideration of the challenges encountered during the data collection process. Some of the challenges included COVID restrictions that affected the availability of respondents, and to some extent, of the data collectors, as well as insufficient reliability of phone calls due to network issues and lack of access to phones by some respondents. Nevertheless, the team was able to overcome all the challenges to complete the datasets and to proceed with data analysis and manuscript writing.
The LAISDAR team is encouraged by the attention that our work is going to garner across various peer-reviewed journals, as the only remaining activity is writing papers and identifying conferences. A monthly workshop is one of the project’s strategies to speed up the publication work to disseminate what the team has achieved so far.
LAISDAR Researchers attending a workshop for manuscript writing in Musanze, Northern Province on 21 February 2023
To promote dialogue around the work we’re doing, LAISDAR will be facilitating a series of events such as general research meetings and dissemination workshops, as well as ensuring our knowledge products are available. To that end, we are also pleased to share the news that one paper has been published in a peer reviewed journal , two papers are under review in peer reviewed journals, and the research team is currently working on three other papers to be finalized and submitted for publication soon.
We would like to express our thanks to Canada's International Development Research Centre (IDRC) and the Swedish International Development Cooperation Agency (Sida) for supporting this initiative through the Global South AI4COVID Program. Additionally, our work would not be possible without the support of the Government of Rwanda and all institutions involved in this consortium.
To know more about this project, please visit our website: Laisdar (rbc.gov.rw) and follow our work on Twitter: @laisdar and Facebook: Laisdar Laisdar.Tags: