Could a machine learning algorithm help predict which trailered boats might be carrying invasive species? A group of researchers, including Rensselaer Polytechnic Institute scientists Sandra Nierzwicki-Bauer and Jeremy Farrell, is developing a computer model that uses data collected during boat inspections to predict whether an uninspected boat is likely to be harboring invasive species. The New York State Department of Environmental Conservation's Invasive Species Grant Program has awarded the project a $78,000 grant.

The researchers report that, when tested against existing data, the first version of the Advanced Response Model for Organism Removal – ARMOR –successfully predicted 141 out of 146 instances in which invasive species were found in a dataset that included over 100,000 boat wash surveys.

“In an actual deployment, if ARMOR had been used to direct inspections, it would have reduced inspection effort by over 50% while sacrificing less than 4% of actual detections,” according to a project summary.

The researchers hope that, by identifying trailered boats with a high risk of carrying invasive species, the algorithm will improve the efficiency and efficacy of detecting and mitigating new invasive species introductions.