Approved Research

Multi-view Investigation of Sleep Disorders Co-Existing with Psychiatric and Neurocognitive Illnesses using Machine Learning

Lay summary

Many people do not sleep well,  disturbed by problems like insomnia. Individuals with sleep problems often have emotional and cognitive complaints and are furthermore  manifested with disorders such as depression, anxiety, and dementia simultaneously or subsequently. These people are in a lower status of life quality and in a higher risk for adverse outcomes.

Hence, revealing the neurobiological causes behind the co-occuring phenomenon of sleep and other mental illnesses, and identifying the bio-markers that allow for early diagnosis and course prediction is critical for better health care and disease prevention. These questions, however, remain underexplored. Previous studies typically focused on single types of data, e.g., clinical rating scales, and the sample sizes are usually small, mostly cross-sectional with no follow-up.

Using the ever-large cohorts of mental health data with both baseline and follow-up assessments collected in UK Biobank, we will investigate 1) the overlaps and differences in biological (e.g., neuroimaging parameters and genetic assays) and behavioral (e.g., clinical ratings and cognitive tests) characteristics among individuals with sleep disorders and different (emotional and cognitive) brain illnesses; 2) how baseline data from individuals with only sleep problems could predict their long-term clinical courses and outcomes.

We will also investigate individuals starting with mental problems but not sleep disorders to reveal the biological roots contributing to the transition from mono emotional/cognitive condition to insomnia-mental/cognitive coexistence in the long-term follow-up.

UK biobank data will therefore be used as herein a wide category of data is covered, including sociodemographics, psychosocial factors, health and medical records, cognitive tests, multi-modal brain MRI, genomics, and other self-reported medical conditions. These data will be combined by machine learning approaches similar to those in our prior work.

This research may guide the development of new treatments for the comorbidity of sleep and other mental disturbances and prevent individuals with early sleep deficits from developing to severe psychiatric and neurological disorders, together leading to better health care and precision medicine.