A breathalyser pregnancy test: How AI is sniffing out the next farming revolution

An electronic nose, powered by artificial intelligence, is helping Dr Bronwyn Darlington and her dairy cows, Soy and Oatley, to fit with the modern tastes in milk names, as Armadale, Flower, Pumpkin and Dixie sniff out the next revolution in global agriculture.Darlington and her Jersey herd are working on technology which promises to modernise farming techniques as old as agriculture itself with the Agscent sensor, like a breathalyser for cattle, which uses nano-sensors to detect chemical compounds in a cow’s breath at concentrations as low as parts per billion.Dr Bronwyn Darlington with her Jersey cows on the family farm in Carwoola, NSW.Alex EllinghausenThe breath analysis device can be used for a range of health-related applications, and a promising one is pregnancy detection, which can be confirmed as soon as 18 days after conception – weeks earlier than an invasive physical exam.This hands-off technique would replace the need for a veterinary test, using elbow-length gloves and a lot of invasive pushing and prodding, which is typically effective only after about 35 days.“Every time you bring an animal in and you do something invasive to it, you create a risk to the person, often a veterinarian, and a risk to the animal. But by doing a breath sample, it’s non-invasive and it gives you a really rich biological sample,” Darlington said.Significant advancements in agricultural productivity came in the “green revolution” of the 1960s, when new crop breeding techniques were deployed to produce high-yielding varieties of rice and wheat to feed the booming global population.More recently, the so-called agtech revolution in the 1990s used Global Positioning System (GPS) technology to boost yields by minutely tracking crops across fields and laser-guiding machinery to sow, till and harvest the field with precision.Agscent uses electrodes coated with nanoparticles to identify a wide range of chemical compounds in an animal’s breath. AI and machine learning “decode” the information and can be trained to recognise the markers for bacteria, viruses and diseases. It can also identify imbalances in diet such as ketosis, which occurs when fat is burned instead of carbohydrates.Katie McRobert, director of the Australian Farm Institute think tank, said there was a deluge of new devices and technology applications promising to boost farmers’ bottom lines, but the problem was finding the time to learn how to use them.Darlington takes a breath sample from Soy, the Jersey cow. Alex Ellinghausen“AI can be a game-changer because the thing that makes farming difficult is the plethora of decisions that everybody has to get through on any given day,” McRobert said.“Australian farmers wear so many hats on any given day – they have to be an agronomist, a soil scientist, a livestock management specialist, an accountant.”Despite these time pressures, technology developers often mistakenly assume that their products will prove popular if they can help farmers boost productivity, even if they take considerable time to learn how to use.This is where the addition of artificial intelligence is coming in.“AI can accelerate decision-making processes when given robust information and good guidelines to work within. It can speed up people’s learning and adoption of technology,” McRobert said.Dr Bronwyn Darlington with her Jersey cows on the family farm in Carwoola, NSW.Alex EllinghausenDarlington said her Agscent product was like a space-age version of timeless technology.“A dog can smell stress in humans or even cancer. A bull will jump through paddock fences to get to a cow.“Our technology simply means we can create a map of that, and if you can create a map of it, the task going forward is how many different productivity benefits can you get.”Cut through the noise of federal politics with news, views and expert analysis. Subscribers can sign up to our weekly Inside Politics newsletter.Mike Foley is the climate and energy correspondent for The Age and The Sydney Morning Herald.Connect via email.From our partners