Patterns: Neural network created that models outcome of various patient therapy methods

A team of scientists from the Ohio State University (USA) have introduced an artificial intelligence (AI)-based program which models treatment options for people with cardiovascular diseases and selects the most effective means of therapy. The respective research was published in the scientific journal Patterns.

The aforesaid team trained the algorithm, named CURE, on millions of people based on de-identified data provided by employers, insurance companies, and medical facilities.

Then the specialists added special information about specific health conditions and treatment methods to the model, focusing attention on strokes. This enabled to get predictions about the effectiveness of different approaches to therapy—and based on the individual characteristics of a person.

According to the creators of this neural network, their CURE outperformed the seven existing analogues in terms of quality of recommendations, and its conclusions were consistent with the results of four clinical studies.

The scientists noted that such an introduction of AI does not lead to the abandonment of clinical research, but enables to save a lot of time and money which are spent during such tests.

Specialists believe that in the future such CURE algorithms will enable to create full "digital replicas" of patients and be a treatment guide for each individual case.

Follow NEWS.am Medicine on Facebook and Twitter