An innovative tool combining Artificial Intelligence (AI) has been introduced by researchers to enhance the accuracy of cause-of-death data in regions with limited resources. This groundbreaking initiative, called the Cause of Death Determination Ascertainment (CODA) project, is funded by the Gates Foundation and spans three years. Its primary objective is to improve mortality data collection in low-income countries, where only eight percent of deaths currently have documented causes.
The CODA project aims to assist health professionals in accurately identifying causes of death in line with WHO standards. Ultimately, this information will help inform public health policies aimed at addressing these critical issues.
According to Philip Setel, vice president for civil registration and vital statistics at Vital Strategies, a global health organization leading the consortium behind this initiative, a significant portion of deaths in Sub-Saharan Africa and various regions of Asia occur outside of health facilities. This poses challenges in determining accurate causes of death.
Setel remarked, “It is an enormous blind spot.” He added, “In Africa, only about one in ten deaths is even registered, as reported by the World Health Organization. The proportion of those that have a reliable cause of death is even smaller. The same issue exists in parts of Asia, where cultural norms favor home deaths, complicating data collection.”
Setel emphasizes that this innovation allows countries to blend AI with established public health methods, such as civil registration and vital statistics systems, to improve the determination of causes of death, even in situations where clinical information is limited. This advancement could significantly enhance understanding of mortality causes and inform prevention strategies.
Data Inputs
The CODA tool utilizes historical data to train its AI algorithms and can be utilized in both clinical and community settings. It has the capability to operate offline, allowing data to be uploaded later when internet access becomes available.
In community settings, information gathered from post-mortem interviews with family members, conducted by community health workers, can be paired with data about the deceased, including age and gender, as well as information on prevalent causes of death in the region, such as malaria.
The CODA tool manages these interviews in real time, dynamically adjusting the suggested questions as the conversation progresses. As the dialogue unfolds, the AI transforms spoken testimonies into structured data, translating described symptoms into terminology that the cause-of-death algorithm can analyze, as explained by Setel.
In health facility environments, physicians familiar with the case can input clinical observations, including patient history, observations, and test results prior to death.
Rather than offering a definitive cause of death, CODA will provide a level of confidence in its recommendations, empowering healthcare professionals to make informed decisions.
Vital Strategies will partner with Northeastern University, the University of Washington, IS Global, RTI International, and the CHAMPS Project Office to conduct limited trials in South Africa and Bangladesh starting in September.
Two university partners are currently working on using historical data to build and test the AI models.
Setel further elaborated that the dataset employed for training the AI consists of rigorously validated deaths in low- and middle-income countries, where causes of death are confirmed through post-mortem investigations. This ensures that the models are trained on high-quality diagnostic standards against which the AI’s performance can be measured.
He acknowledged that while autopsies and post-mortem examinations in health facilities yield high-quality data, they are often costly and limited in scale.
Informing Policy
Beyond its clinical applications, the CODA initiative aims to empower governments in their allocation of health resources.
Mary-Ann Etiebet, president and CEO of Vital Strategies, emphasized the importance of accurate cause-of-death data for effective policymaking. She pointed out, “Less than three percent of global health financing is allocated to non-communicable diseases, despite their significant burden.”
“A lack of precise data obstructs policymakers’ abilities to allocate resources effectively,” she added. Etiebet also mentioned the ethical implications involved in sensitively interviewing families about deceased loved ones, along with the legal complexities surrounding AI’s status within existing legal frameworks.
To address these concerns, a scientific advisory committee will be established, ensuring that the tool is ethical, legal, and culturally sensitive. Laura Ferguson, director of research at the University of Southern California’s Institute on Inequalities in Global Health, noted the ongoing efforts related to digitalizing civil registration and vital statistics.
Ferguson remarked, “We must start with the problem—insufficient data regarding who is dying, where, and why—and evaluate what tools could be beneficial.” She asserted that AI has the potential to enhance understanding of mortality causes, enabling governments to allocate resources to evidence-based interventions aimed at reducing mortality rates effective.
Ultimately, Ferguson stressed the significance of implementing safeguards like transparency, data protection, government involvement, and collaboration with end users. “AI can be a valuable resource, but its deployment is key to its practical effectiveness.”
This piece was produced by SciDev.Net’s Global desk.
