Policymakers in Ghana have recently indicated that tackling unhealthy food production and improving food processing would be the most likely actions to address the associated effects of unhealthy foods, such as obesity and Non-Communicable Disease risk factors, especially in children and adolescents.
African households increasingly rely on unhealthy, ultra-processed food from the formal food and retail system. Simultaneously, many Sub-Saharan countries' food production and processing environment is deteriorating. More attention is being given to quality requirements for food. An important stepping stone for monitoring and guaranteeing quality is tracking all elements through the complete production chain, from the base material overall processing steps to the final wholesale and retail. Dr Eric Tutu Tchao is thus researching the development of an Open Toolbox for safe food monitoring (DOTbox).
Policymakers in Ghana have recently indicated that tackling unhealthy food production and improving food processing would be the most likely actions to address the associated effects of unhealthy foods, such as obesity and non-communicable disease (NCD) risk factors, especially in children and adolescents. The same policymakers acknowledge that food environment-related responses to NCD prevention are hampered by a lack of data from the value chain for decision-making and appropriate technological interventions and monitoring in Ghana. Government agencies or sales channels often do monitoring, often with proprietary and closed solutions.
Dr Tchao is of the opinion that blockchains could solve this problem by providing invariable information about the entire production and transportation line for groceries, including ingredients and their origins. Internet of Things platforms cover the complete information flow, from the sensor to the cloud, including identification, authentication, storage, communication, and data analytics. Blockchain technology enables the secure transaction of information across company boundaries. In particular, it allows the implementation of immutability, where each party involved is directly connected to the blockchain via a computer. Traceability can be guaranteed.
Dr Tchao has, as part of his research, reviewed existing traceability solutions currently being used in the industry. The review results formed the basis for him to develop a new Blockchain Architecture for a resource-constrained environment like Ghana. Last year, the review and architecture development outcomes were presented at a Computational Science and Computational Intelligence conference in Las Vegas. He has developed a GitHub project to publish the Blockchain implementation codes. After all security enables are guaranteed, the project would be opened to the public. Dr Tchao has also developed an Anomaly Detection model called AutoSafe, a Deep Reinforcement Learning (DRL) model. The AutoSafe model was presented at the AfricaAI conference in Kigali.
Datasets for Palm Oil Adulteration are being generated - a first of its kind and will be made openly available when the data-gathering process is completed. For this, he engaged local market women and oil producers in the study areas of Accra and Kumasi. These engagements, coupled with those of Food Scientists at the Kwame Nkrumah University of Science and Technology (KNUST) in Ghana, helped identify the various contaminants and palm oil types on the markets. Ghana's Food and Drugs Administration made inputs during the needs requirement analysis phase. Their inputs were significant in determining the parameters to detect palm oil anomalies and adulteration.
Dr Tchao says his mentor, Prof Lise Korsten, provided valuable insights into how to gather relevant data for the AutoSafe Model training. "She indicated that my current solution would only punish the market women and not the other players in the supply chain. She proposed that I undertake further data gathering in the supply chain to track the adulteration's source. She also said the toolbox should be developed to enable the crowdsourcing of data, where the app's users would also contribute to the existing datasets. These precious inputs are the basis for the extended data-gathering phase." His supervisor, Prof Jerry John Kponyo, played a vital role in developing and optimising the AutoSafe Model. He provided funding, enabling him to present the AutoSafe Model at the AfricaAI conference in Kigali. "My supervisor reviewed all manuscripts and provided critical feedback to enhance the quality of the research work."
Heidi Sonnekus | FAR-LeaF programme
