“..we each are more likely to die by our own hand than by someone else’s” – WHO, 2016
From the paper, “Digital Phenotyping of Suicidal Thoughts.”
9% of adults worldwide have seriously considered suicide at some point in their lives. It’s a worrying statistic, with one person committing suicide every 40 seconds and for each suicide there are 20 attempted suicides. Theoretical research has suggested that there could be two types of suicidal thinking:
- One type having great fluctuations in severity to suicide thinking.
- The other type having consistent levels of suicidal thoughts without fluctuations.
There are thoughts that the increased prevalence of the internet and social media with new technologies could be a reason for increasing suicide statistics (1). Despite this, researchers have looked into ways of using technology to predict our mental health. One way to do this is to classify suicidal thinking into different subtypes. This would be extremely useful as a different phenotype or class of suicidal thinking could have different predictors/symptoms, different treatments and different responses to some treatments. Two sample groups were used in this investigation. One sample contained unhospitalised people who had attempted suicide within the last year and the second were current hospitalised individuals for a recent suicide attempt or for severe suicidal thoughts.
How Was Suicidal Thinking Investigated?
Everyone in the sample groups had a smartphone that would send them 4 prompts a day at random predefined intervals. The individual would chose whether or not to answer the prompt or not, which would then be sent to researcher for analysis and monitoring.
At each of the four prompts, the participants were asked three questions:
- How intense is your desire to kill yourself right now?
- How strong is your intention to kill yourself by suicide right now?
- How strong is your ability to resist the urge to kill yourself by suicide right now?
Each question would have a continuous score from either 0 to 4 or 0 to 9 as possible answers, with 0 indicating not strong and 4/9 indicating very strong. The answer given would be called the ‘score’.
An analytic strategy called ‘latent profile analysis’ or ‘LPA’ was used to generate profiles of suicidal thinking based upon questions with an infinite amount of answers. LPA in this experiment used 5 different indicators to measure suicidal thought patterns.
The RMSSD is a method used to assess variability between two of the same data points at different times, such as heart rate from one day to the next. If there is a large value for RMSSF, then there is more variability from one time point to the next, showing a jagged pattern. Here, for suicidal thinking, the higher the RMSSD the more variable suicidal thinking prevalence is.
Five different categories of suicidal thinking were determined based upon the mean scores and the variability of the mean scores. A low mean score means that individual they have fewer suicidal thoughts than someone with a high mean. Variability is used to show if an individual has fluctuations in terms of suicidal thinking. Someone can generally have low suicidal thinking but can go through periods of severe suicidal thoughts. The categories determined are:
- Low mean score with low variability
- Low mean score with moderate variability
- Moderate mean score with high variability
- High mean score with low variability
- High mean score with high variability
Those placed within categories 4 and 5 could be seen as high risk to attempt suicide with categories 1 and 2 being low risk. Within this experiment, those who had attempted suicide a month before the study were placed within category 4 based upon the predictors and the prompts. Therefore, someone with a high mean score (so high amount) of suicidal thinking with little fluctuations in thinking were characteristic of someone who had recently attempted suicide.
Being able to categorise types of suicidal thinking could prove extremely useful. It can lead to development of treatments that are specific to a type of thought process which could prove more effective than one generalised treatment strategy. It could also be used to spot the likelihood of someone attempting suicide after trauma involving medical care or even in a G.P. setting.
However, predicting and categorising suicidal thinking is more difficult than assigning data points to it. Thought processes are, obviously, different from person to person. More research needs to take place to see whether categorising people’s suicidal thinking is reliable and whether it can be used to develop different treatments. Time will tell whether research papers like this one can be used in a clinical setting to prevent suicide attempts.
— By Daniel Baird
(1) Luxton, David D., Jennifer D. June, and Jonathan M. Fairall. “Social Media and Suicide: A Public Health Perspective.” American Journal of Public Health 102, no. S2 (May 2012): S195–200. https://doi.org/10.2105/AJPH.2011.300608.