The Center for Data Innovation recently spoke with Tayab Soomro, co-founder of PathoScan, a Canadian company that helps farmers identify crop diseases using AI. Soomro explained how Pathoscan’s portable device, the PathoBox, allows users to detect plant pathogens directly in the field, providing results in minutes, and helping farmers efficiently protect their crops.
David Kertai: What agricultural problem does Pathoscan address?
Tayab Soomro: PathoScan addresses the challenge of slow and inaccessible crop disease testing. Traditional lab methods require farmers to send plant samples away and wait several days for results, by which time a disease may have already spread too far to control effectively. These delays often force farmers to rely on guesswork or apply pesticides broadly across their fields. PathoScan solves this problem by enabling fast, on-site detection, allowing farmers to identify issues early and take targeted action only where it’s needed.
Kertai: How does Pathoscan’s device apply AI to detect disease on crops?
Soomro: Our device, called the PathoBox, combines molecular testing with an AI model to deliver quick and accurate results in the field. It uses a process called loop-mediated isothermal amplification, which works like a photocopier for DNA or RNA, rapidly making millions of copies of any pathogen’s genetic material in a plant sample. The built-in AI model then examines the data, identifying the pathogen and estimating the infection’s severity, helping farmers decide on the most effective next steps.
Kertai: What results does PathoScan deliver to the user?
Soomro: After a test, the farmer receives a clear, easy-to-understand result indicating whether the target pathogen is present in the sample. The device also provides an infection severity rating, high, medium, or low, that estimates how widespread or advanced the disease is. Because results are available in about 30 minutes, farmers can respond immediately, focusing treatment only where needed and preventing further spread.
Kertai: How do you ensure the accuracy and reliability of your AI model?
Soomro: We verify PathoScan’s accuracy by comparing its results to standard laboratory tests. In our field trials, we test one plant sample using the PathoBox and send a paired sample to a reference lab, which performs traditional diagnostic tests, such as polymerase chain reaction or DNA sequencing. We then compare results using key performance metrics like sensitivity, how well the test detects true positives, and specificity, how well it avoids false positives. To ensure the device remains reliable beyond controlled conditions, we also evaluate its performance in real-world farm settings, including remote areas with limited internet access.
Kertai: Can you share any real-world use cases of your device in the field?
Soomro: We’ve carried out several field trials in Saskatchewan, Canada, with both agronomists and greenhouse growers. In one greenhouse, we used the PathoBox to test dozens of plants and detected an early-stage Botrytis infection, a common fungal disease, within about 30 minutes. Thanks to the quick results, the grower was able to immediately isolate affected plants, stop the infection from spreading, and avoid unnecessary pesticide use.
