AutMedAI Decoding Autism in Toddlers
Researchers at Karolinska Institutet in Sweden have developed a new machine learning, AutMedAI, to predictive ability of autism in young children with 80% accuracy using minimal information.
According to the study entitled “Machine Learning Prediction of Autism Spectrum Disorder From a Minimal Set of Medical and Background Information” and published in JAMA Network Open, this new AI model could improve the early detection of autism, making it easier to provide early support to children with autism as soon as possible.
Smiles and Words Detect Autism
To train the AI model, the team used a large database from the US called SPARK, containing information on approximately 30,000 individuals with and without autism spectrum disorders.
The researchers chose 28 different parameters related to the age, first smiles or spoken words, first short sentence, and the presence of eating difficulties.
Using these parameters, the team developed four totally different machine-learning models to detect patterns in the data. These parameters are easily available and thus do not require long assessment or any medical tests, making the model practical for early diagnosis.
AutMedAI was tested on nearly 12,000 children and successfully identified about 80% of those with autism. The high accuracy of this AI model could significantly impact autism diagnosis, allowing for early intervention, which can, in turn, enhance children’s development.
A promising Advancement
Health professionals see the success of AutMedAI as a game changer, given that with such disorders early detection is often a challenge.
Commenting on the outcomes of the study, Shyam Rajagopalan, the study’s first author said, “The results of the study are significant because they show that it is possible to identify individuals who are likely to have autism from relatively limited and readily available information.”
Looking ahead, the research team plans to further validate and refine the AI model in clinical settings. They are also exploring the integration of additional genetic information, which could further enhance its accuracy and predictive ability of autism.
Final Thoughts
AutMedAI represents major progress in the diagnosis of autism. Currently, there is no specific or permanent treatment for Autism, so early detection is most necessary.
The AI model offers parents hope by its predictive ability of identifying autism symptoms with just minimal data, achieving an accuracy rate of 80%. Its high precision can result in timely interventions that benefit much in children’s development.
Inside Telecom provides you with an extensive list of content covering all aspects of the Tech industry. Keep an eye on our Medtech section to stay informed and updated with our daily articles.