Brain Connectivity Patterns a Reliable Indication of Suicide Risk in Elderly Patients Suffering From Depression: HKU

Diagram sequence to explain the mechanism from research findings. Courtesy of HKU

As the incidence of depression in the elderly continues to rise, the risk of death caused by ideation, plans, and behavior toward suicide among the elderly is likely to be higher than that among other age groups. A study from HKU’s State Key Laboratory of Brain and Cognitive Sciences found that brain connection patterns can help predict suicide risk in elderly patients with depression. The findings have been published in Nature Mental Health, a sister journal of Nature.

A research team led by Tatia Lee Mei-chun, director of HKU’s State Key Laboratory of Brain and Cognitive Sciences, used brain imaging data and advanced computer algorithms to predict the severity of suicide risk in patients with depression in later life. The team used a predictive model based on connectomics to predict through whole-brain resting-state functional connectivity (brain active connections when not performing any specific cognitive tasks) and white matter structural connectivity (structural connections between brain regions) data to predict suicide risk.

Improve Forecast Accuracy

The study recruited 91 elderly patients with depression, including 37 who had no suicide tendencies, 24 who had suicidal thoughts or plans, and 30 who had attempted suicide for evaluation. The researchers used advanced computer algorithms based on brain connection characteristics and machine learning to classify the three groups of patients.

The results show that compared with using only questionnaire scores for assessment, using brain function and structural connectivity as learning features can improve the accuracy of predictions and identifying depression with a higher risk of suicide in the two independent data samples.

Hope to Design Better Personalized Screening

Ms. Lee pointed out that when distinguishing elderly patients with depression who are at higher and lower suicide risks, functional connection and structural connection characteristics can improve the accuracy of classification. The team pointed out that multimodal brain connections can help capture individual differences in suicide risk in elderly patients with depression and are expected to identify further patients needing in-depth evaluation and intervention. In the future, the team hopes to collect patients’ brain imaging, psychological and behavioral data, and integrate them into predictive models to establish personalized suicide risk scores and better design personalized screening and intervention plans.