AI-Powered AMR Detection System in Ethiopia: A One-Health Solution

AI-powered solution aims to tackle the pressing issue of AMR in Ethiopia through a comprehensive one-health approach. By improving AMR detection and monitoring systems at primary level of care settings, the solution seeks to mitigate the impact of AMR on health and development.

Berhanu Seyoum Endale (PhD), the Primary Investigator of the solution, will be the Team Lead. 

AMR is a growing global health concern. It was a direct cause of 1.27 million deaths of the 4.95 million drug-resistant infections in 2019. It also leads to increased higher healthcare costs. If appropriate intervention is not designed to mitigate this crisis, AMR could force up to 24 million people into extreme poverty by 2030 globally. AMR not only affects human health but also impacts agriculture and livestock production.

Ethiopia is one of the African countries that struggle with limited resources, particularly in trained workforce and adequate laboratory infrastructure. As a consequence, the burden of AMR in the country is huge. In Ethiopia, AMR-attributable deaths accounted for 21,200 deaths and 85,300 deaths directly associated with AMR in 2019. Therefore, our solution will support early detection and response to AMR. The solution is affordable, appropriate, and easy to use at the point of care in the Ethiopian context. Large data sets from humans, animals, and the environment at the community level, will be analyzed by AI detection systems to find patterns that might point to the presence of AMR. Consequently, AMR strains can be swiftly and effectively detected, allowing for enhanced detection and informed decision-making towards appropriate anti-microbial use.  

Our solution primarily serves for human and animal healthcare providers at the primary level of care, researchers, and public and animal health policy makers dedicated to combating antimicrobial resistance (AMR). The proposed new AI-powered detection system for AMR in Ethiopia will improve early detection, design targeted intervention, support evidence-based decision-making on rational use of antimicrobial agents, enhance capacity building, contribute to global health security efforts in combating AMR both in human and animal sectors, and foster regional collaborations. 

To understand their needs, we actively engage in collaborative partnerships, conduct user surveys, and gather feedback from these stakeholders throughout the development process. This iterative approach ensures that our AI-powered AMR Detection System aligns with the real-world challenges faced by those on the front lines of combating AMR. By prioritizing user insights and fostering continuous communication, we aim to provide a solution that is not only technologically advanced but also directly addresses the practical needs of our targeted end users. 

The AI-powered AMR detection system contributes vital public goods aligned with the One Health approach:

1. Improved AMR Detection and Rational Use of Antimicrobial: The solution improved the detection of the 19 prioritized bacterial pathogens in One Health. Besides, rational use of  antimicrobial agents at primary level of care will be  increased, which will resulted in for reduction of resistant bacterial strains.
2. Integrated Surveillance: Integrating into the databases of already-existing One Health systems enables thorough monitoring and decision-making, promoting coordinated responses between the human and animal health sectors.
3. Informed Decision Making: The solution enables veterinarians and healthcare providers to make the right decisions about antimicrobial treatment by providing them access to up to date information, thereby promoting efficient primary level of healthcare services.
4. Capacity Building: Technical capacity building guarantees that healthcare providers utilize the solution effectively, improving AMR detection and surveillance systems.
5. Scalability and Sustainability: The solution, designed to tackles evolving challenges in AMR detection and rational use of antimicrobial, is built for sustainability and scalability, guaranteeing long-term impact.

Hence, our AI-powered AMR detection solution improves national surveillance systems, facilitates informed decision-making, and boosts the capacity of the veterinary and healthcare providers to effectively combat AMR.

Our AI-powered AMR detection solution is designed to have a significant impact on the target population by addressing the global threat of AMR. Here's a simplified explanation of how our solution to make a difference:

1. Antibiotic Prescribing and Rational Use Patterns: We will monitor changes in antibiotic prescribing and patterns following the implementation of our solution. Specifically, we will assess the proportion of appropriate antibiotic prescriptions and rational use. 
2. Targeting Underserved Populations: Our solution can be deployed in various primary healthcare settings, including remote and resource-constrained areas, to extend access to accurate AMR detection.  
3. Scalability and Accessibility: Our solution is designed to be scalable and accessible, leveraging digital technologies to reach a wide range of healthcare settings and populations. By integrating our solution into existing healthcare infrastructure and leveraging mobile platforms, we can extend its reach to remote and underserved areas where AMR surveillance is particularly critical.

The development of an AI-powered AMR detection system will pass through the following process:

Year 1: Laying Foundation:

• Data Preparation
• Algorithm development
• Software integration and user interface design
• Database integration
• Testing and validation
• Conduct project kick-off workshop,
• Conduct baseline situation analysis,
• Stakeholder engagement, collaboration and partnership

Year 2: Model validation and piloting

• Testing and validation of the model
• Pilot testing of AI model
• Design and implement training programs.
• Establish a system for continuous improvement.
• Document and disseminate learning.
• Stakeholder engagement, collaboration and partnership
• Conduct consultative, validation, and approval workshops.
• Conduct midterm project review workshop.

Year 3: Integration and scale-up

• Establishment of an AI-powered AMR user support system
• Conduct capacity-building activities
• Advocate and integrate the AI Model system with the national AMR database system.
• Scale up pilot projects and Optimization

Mainstreaming Strategies:

• Focus on equity and inclusion: Ensure our solution benefits all populations, addressing equity issues (access and affordability barriers).
• Sustainability plan and strategy 
• Continuous Learning and Adaptation: Continuously monitor and evaluate, adapting and improving based on feedback and data.

Monitoring and evaluating the impact of our AI-powered AMR detection system is crucial for assessing its effectiveness and guiding continuous improvement. Here's our plan to monitor and evaluate impact, along with specific, measurable indicators:

Antibiotic Prescribing Patterns: We will monitor changes in antibiotic prescribing patterns following the implementation of our solution. Specifically, we will assess the proportion of appropriate antibiotic prescriptions based on the detected AMR profiles.  

Patient Outcomes: We will evaluate the impact of our solution on patient outcomes, including treatment success rates, length of hospital stays, and incidence of AMR-related complications. By tracking these indicators, we can assess the clinical benefits of our solution in improving patient care and reducing the burden of antimicrobial-resistant infections.  

Healthcare Cost Savings: We will estimate the cost savings associated with the implementation of our solution, including reductions in unnecessary antibiotic use, and healthcare-associated infections.  

By monitoring these specific, measurable indicators, we can assess the impact of our AI-powered AMR detection system on antibiotic prescribing patterns, patient outcomes, and healthcare cost savings. These indicators provide valuable insights into the effectiveness and value proposition of our solution, guiding further refinements and scale-up efforts to address the global challenge of AMR.

Availability of comprehensive and quality data on AMR from humans, animals, and the environment including healthcare facilities, and veterinary clinics, can be challenging.  It can also be difficult to integrate data from different sources, such as national and regional surveillance systems, laboratory databases, and electronic health and veterinary records. For an integrated One Health solution to be successful, these systems must be interoperable and seamlessly integrated.

A multidisciplinary approach involving experts from the fields of healthcare, veterinary medicine, data science, policymakers, and ethics will be employed to address the aforementioned barriers. In addition, various stakeholders will be engaged in the work to overcome the mentioned challenges and accomplish the objectives of deploying an AI-powered AMR detection system as a sustainable integrated One Health solution.

Financial barriers can also pose significant challenges to the widespread adoption and effectiveness of the AI-powered AMR detection system. We planned to secure a budget through public-private partnerships, grants, and cost-sharing arrangements to ensure equitable access to the AI-driven AMR detection system and mitigate the economic burden on stakeholders.

Antimicrobial Resistance (AMR) is becoming a global public health challenge and is considered as a "silent tsunami”. The situation needs much attention in developing countries where the AMR detection system is very poor due to limited resources to establish laboratory infrastructures and to deploy trained and skilled healthcare providers in the primary healthcare setting. Human resources for primary healthcare is covered by less skilled professionals, mainly in the form of task-shifting.

We are applying to The Trinity Challenge because we recognize the critical importance of collaborating with diverse stakeholders to address global health challenges such as AMR through an AI-powered AMR detection system at community level. 

The Trinity Challenge provides a unique platform for us to overcome key barriers in advancing our AI solution. By participating in the challenge, we can leverage collaborative networks, expertise, and resources to realize our solution, ultimately contributing to addressing the critical global challenge, AMR.

• Ethiopian food and drug Authority
• Ethiopian Agricultural Authority
• Ethiopian Artificial Intelligence Institute
• Armauer Hansen Research Institute