Artificial intelligence machine learning technologies significantly enhance research potential, provided that the data is good. Neural networks that mimic human learning, pattern recognition, and data analysis plow through massive amounts of data at astonishing speeds with vast implication potentials for medical research. We’ve written previously about the machine learning used to predict pain levels from video clips, uncover hidden diabetes insights, and even target specific health conditions such as Huntington disease. We also mentioned a Vanderbilt University study in which machine learning successfully predicted suicide with 80% accuracy from analyzing large numbers of health records.
Researchers at Carnegie Mellon‘s Center for Cognitive Brain Imaging more recently published a study in Nature Human Behavior in which machine learning algorithms analyzed MRI brain scans to detect neural representations of suicidal ideation in adolescents. The research question was whether words like “death” and “trouble” produced specific neural signatures in the brains of adolescents with high suicidal risk. Overall the study found the machine learning analysis identified suicidal thinkers with 91% accuracy with brain scans from 17 suicidal adolescents and 17 controls. Further, among the suicidal ideators, the artificial intelligence algorithms correctly identified those who had attempted suicide with 94% accuracy.
Brain research is proceeding quickly on many fronts. Studies such as this one show that early identification and intervention may be possible for behavioral problems, and not just disease.