Suicide in the United States is increasing every year making it the 10th leading cause of death among Americans. Healthcare experts have the right tools to help people who are suicidal or on the verge of committing suicide. The main challenge, most of the time, is they don't know who needs the help most. However, all this could change with the help of artificial intelligence.
According to the statistics released by the United States Center for Disease Control (CDC), suicide rates have been increasing from 1999 to 2014. The highest is among middle-aged women which increased by 63 percent over the period, while the overall suicide rate increased by 24 percent.
Health care providers often rely on well-known risk factors, such as drug use or depression, to predict if the person has suicidal tendencies. However, suicide is more than just these factors but a combination of inter-related life events that are very complex.
In order to decode this complexity and prevent suicidal people from committing it, medical professionals are turning to artificial intelligence for help because it can understand these complexities much better and map out the risk factors that lead to suicide.
A team of researchers from Florida State University has conducted an experiment using artificial intelligence and machine learning to predict suicide before it happens. They fed an algorithm the health records of more than 3,000 people who had tried to commit suicide.
The machine analyzed different patterns using different factors that led to a suicide attempt. Some of these factors include the medications the people took as well as the number of ER trips they made over the years. The algorithm also discovered some unusual factors as it continued learning the data it was fed.
Jessica Ribeiro, a psychologist and part of the research team, said the artificial intelligence was able to accurately predict that the person will commit suicide within the next two years. More so, it was 92 percent accurate in predicting a suicide attempt within the next week.
The study was published in the journal Clinical Psychological Science.
