Children with autism often show differences in speech compared to children with normo-typical development, both in speed, pitch, intonation and rhythm. These differences have been very difficult to characterise consistently and objectively for decades. So we ask ourselves: Are speech characteristics really a useful tool for diagnosing the disease, and can Machine Learning help in this task?
According to a study by Northwestern University, Machine Learning models can help diagnose autism by identifying speech patterns in different languages.
Read the article 👉 AI Detects Autism Speech Patterns Across Different Languages 👈
This study has used machine learning to identify speech patterns in children with autism in English and Cantonese, providing data that could help scientists better understand the origin of the disease and develop new therapies.
All languages are structurally different. But finding similarities between the speech patterns observed in autism in languages as different as English and Cantonese is especially significant. This fact indicates that these speech traits must be strongly influenced by the genetic liability of autism.
No doubt this kind of predictive models can facilitate the diagnosis of autism helping to reduce the burden on health professionals, contributing to improving the understanding of this condition, and providing a valuable tool that could be used for the diagnosis of other types of disorders or diseases.
This is just one of an infinite number of examples of the application of Machine Learning or other predictive models. Prediction is a key and valuable capability that helps companies and societies around the world to make better decisions. It can be used for example to predict demand, identify fraudulent activities, detect equipment breakdowns before they occur, or anticipate potential risks. A real gold mine to exploit.
Want to learn more about predictive modeling and machine learning?