Disrupting psychiatry with machine learning algorithms

     CHALLANGE Autism Spectrum Disorder (ASD) affects 1 in 59 children, that’s over 10 million in Europe and the US, with costs to society 5 times as high as stroke or hypertension.

Current diagnosis of Autism Spectrum Disorder relies on parent behavioral questionnaires and semi-structured interactive evaluations of ASD symptoms. The clinical assessment is fully dependent on the expertise of the clinician. After diagnosis, there is currently no method of identifying patients for which certain therapies may work.

At the moment, no drug or other treatments are approved that actually  treat the core symptoms of ASD: social and language deficits and restricted and repetitive behaviors. This is because there is generally poor insight into the underlying biological mechanisms of ASD.

     OPPORTUNITY There is a huge opportunity to provide accurate measurement of the underlying biological mechanisms of ASD and mental health issues in general. Shifting from an subjective, symptoms based endpoint towards an objective, physiological based endpoint will change the lives of people living with a mental health issue.

     SOLUTION By combining EEG with Artificial Intelligence, Aspect Neuroprofiles is developing a method to objectively measure and monitor the physiological processes in the brain that are related to ASD. Aspect Neuroprofiles uses the latest theories in neuroscience with advanced machine learning techniques to analyse EEG signals. By analysing imbalances in the excitation and inhibition of neuronal firing, it is possible not only to objectively identify a disorder that until now was solely based on subjective observation of symptoms, but also predict the effectiveness of its treatment. Furthermore, this method could greatly contribute to drug discovery and drug repurposing.

Unique Selling Points

  • Innovative objective measurements, redefining the way we view disorders such as Autism.
  • A new way to stratify participants in research and drug discovery and repositioning
  • Advanced machine learning algorithms, finding and combining biomarkers for improved results
  • Clinical validation of the method

     STATUS The method has been validated in 263 patients. Patent application filed in Q1 2018. The machine learning algorithms are protected by copyrights.