AI Expert System Diagnoses Crop Pests Accurately

Crop diseases and pests cause massive losses in agriculture, threatening global food supplies and farmer livelihoods. In many regions, expert advice is scarce when needed most, leading to delayed responses and reduced yields. This research addresses that gap by developing an AI-based expert system that provides rapid, reliable diagnoses for common pests affecting key crops like rice, tobacco, tomato, pepper, corn, cucumber, and bean. The system empowers farmers to take immediate action, enhancing productivity and supporting sustainable agriculture goals.

The key finding is that this expert system can accurately identify specific pests and diseases by analyzing symptoms input by users. For example, it diagnoses issues such as Pyricularia oryzae in rice, Tuta absoluta in tomatoes, and Botrytis cinerea in peppers, offering both scientific and common names for each problem. This capability mimics the decision-making of a human specialist, ensuring that even non-experts can obtain precise diagnoses without prior knowledge.

Methodologically, the researchers built the system using SWI-Prolog, an open-source programming language ideal for handling rule-based knowledge. They structured the system into three core components: a knowledge base containing expert information on pests, an inference engine that processes user inputs to reach conclusions, and a user interface that guides farmers through simple yes-or-no questions. Development involved five phases: preliminary research on expert systems, planning, knowledge definition from literature and specialist interviews, coding and verification, and final evaluation to ensure error-free operation.

Results show that the system successfully diagnoses a wide range of pests, as detailed in tables and figures from the paper. For instance, Figure 8 illustrates a questionnaire for identifying Pyricularia oryzae in rice, and Figure 9 displays the diagnosis outcome with brief explanations. The interface, shown in Figures 5-7, is user-friendly, allowing farmers to interact seamlessly and receive treatment recommendations. Testing confirmed that the system operates reliably, providing quick responses that help prevent crop damage.

In context, this innovation matters because it supports food security by reducing agricultural losses, which are a major concern in achieving UN Sustainable Development Goal 2 to end hunger. Farmers in areas like Consolación Sur, Cuba, where the study was focused, can use the tool to make informed decisions, potentially increasing yields and income. By democratizing access to expert knowledge, the system helps small-scale farmers compete more effectively and adopt sustainable practices.

Limitations noted in the paper include the system's reliance on predefined knowledge, meaning it may not cover emerging or rare pests without updates. Additionally, it requires user input accuracy; incorrect symptom descriptions could lead to misdiagnoses. Future work could expand the knowledge base and integrate real-time data for broader applicability.