QSAR Classification Modeling Using Machine Learning with a Consensus-Based Approach for Multivariate Chemical Hazard End Points

Description

A team of researchers build a computer method to predict whether chemicals are harmful, in a 2024 study whose head byline begins with Yunendah Fuadah. Testing chemicals on people or animals is slow and raises ethical concerns, so models that sort chemicals by risk are widely sought. The study covers eight harm outcomes, from heart harm to skin and eye irritation.

The team describe each chemical with fingerprints, key fragments, computed traits, and basic physical values. For each kind of descriptor, they pick the best of three learning methods and combine the winners into one consensus model. Scores for telling harmful from safe chemicals ranged from fair to strong across the eight outcomes. The authors present the frame as a reliable, ethical aid for regulators and labs that screen new chemicals.

Catalog record

Source: https://europepmc.org/articles/PMC11683616