Machine learning, a type of artificial intelligence that provides computers with the ability to learn without being explicitly programmed, could allow clinicians to predict who will attempt suicide up to two years in advance with 80 percent accuracy, according to a new study.
The study, led by Florida State University researcher Jessica Ribeiro, used algorithms to predict with 80 to 90 percent accuracy whether someone will attempt suicide as far off as two years into the future.
That accuracy rate increases even further as a person's suicide attempt gets closer. When artificial intelligence focuses on general hospital patients one week before a suicide attempt, for example, the accuracy rate increases to 92 percent.
"This study provides evidence that we can predict suicide attempts accurately," Ribeiro said in a statement to FSU News. "We can predict them accurately over time, but we're best at predicting them closer to the event. We also know, based on this study, that risk factors — how they work and how important they are — also change over time."
The Centers for Disease Control and Prevention describes suicide as a "significant problem in the United States." It ranked as the 10th leading cause of death in the country in 2014 with nearly 43,000 suicides that year. A breakdown of that data showed that among the 10 to 34-year-old cohort, suicide ranked above homicide as the second leading cause of death. Among those 35 to 54, suicide ranked as the fourth leading cause of death. Among adults 55 to 64 suicides ranked as the eighth leading cause of death. It did not feature in the top 10 causes of death among adults older than 64. It is also estimated that suicides result in $44.6 billion in combined medical and work loss costs.
Ribeiro's paper, titled "Predicting Risk of Suicide Attempts Over Time Through Machine Learning," will be published by the journal Clinical Psychological Science, according to FSU News.
Ribeiro recently worked on a study by FSU psychology assistant professor Joseph Franklin which showed that 50 years of suicide prediction research did not show any real progress. The traditional risk factors identified over the past half century to predict suicidal behavior yielded results no better than random guessing, said the report. This latest study using machine learning is a major breakthrough.
"It was really sad," Ribeiro said of the Franklin study. "Fifty years of research with really smart people working on this and no real change. We can see that in the suicide rates. I'm not saying machine learning is the panacea, but these kinds of techniques and changes in the status quo can really disrupt a stagnant research area."
Working from the findings of that study, Ribeiro teamed up with Franklin and Colin Walsh of Vanderbilt University Medical Center and accessed a massive data repository containing the anonymous electronic health records of about 2 million patients in Tennessee. It was the largest project of its kind in history.
They identified more than 3,200 people who had attempted suicide and then used algorithms to examine their medical records and "learn" which combination of factors in the records could most accurately predict future suicide attempts.
"The machine learns the optimal combination of risk factors," Ribeiro explained. "What really matters is how this algorithm and these variables interact with one another as a whole. This kind of work lets us apply algorithms that can consider hundreds of data points in someone's medical record and potentially reduce them to clinically meaningful information."