Predicting mood states with sleep data: Advances in mental health care

Find out how analyzing simple sleep patterns can revolutionize psychotherapy by accurately predicting mood states.

Study: Accurately predicting mood events in bipolar disorder patients using wearable sleep and circadian rhythm components. Image credit: Chay_Tee / Shutterstock

A study published in a journal NPJ Digital Medicine describes the development of mathematical methods that can accurately predict future mood events using only the sleep history and past mood states of patients with mood swings .

Rear end

Sleep disturbances are closely related to various mood disorders, including major depression and bipolar disorder. Monitoring sleep patterns related to mood and performance often uses wearable devices and smartphone sensors that collect physical and behavioral data from patients in real-life situations.

Previous studies using physiological data from wearable devices to develop machine learning methods have shown promising results in identifying people at risk of depression. A combination of machine learning and wearable technology has been used to predict daily emotional states and manage emotional states in patients with mental disorders. However, these models require many types of data, including sleep, heart rate, brightness, phone usage and GPS, which prevents their use in real life.

In this study, scientists have developed a mathematical model for predicting emotional states that only requires the patient’s history of sleep patterns and previous emotional episodes.

Model Development

Scientists collected sleep pattern data and past mood data from 168 patients with major depression or bipolar disorder, aged 18-35 , and ethnic Koreans. Sleep data included complete sleep records for at least 30 days.

They processed sleep-wake pattern data and found a total of 36 sleep-wake patterns and circadian rhythms, which were used as inputs for a machine learning algorithm that was intended to predict seasonal events. coming of depression, manic, and hypomanic in patients with mood disorders. By analyzing the importance of these components, they found strong relationships between sleep, circadian rhythms, and emotional states.

Key features included circadian phase and amplitude Z-scores as well as wake time during long sleep windows, which emerged as the most important predictors of emotional events.

Model Verification

Scientists have confirmed the effectiveness of this model by using previous patterns of patients’ wakefulness and circadian rhythms. They chose a certain type of 60 days for each patient, half of which represent episodic days. They included data from this list in the training set of the model and used the following data from the training list for validation.

Using training data from 60 days, the model accurately predicted next-day depressive, manic, and hypomanic episodes, with AUC (Area Under the Curve) values ​​of 0.80, 0.98, and 0.95, respectively. The level of accuracy in predicting manic and hypomanic episodes remained high when more than 30 days of training were used. However, with reduced training data, the accuracy of the depression event model is greatly reduced. Because of these expectations, scientists have said that sufficient training data is needed for the model to maintain its high accuracy.

This study identified difficulties in predicting hypomanic episodes due to possible non-monotonic relationships between circadian phase and mood states. This difference highlights the need for further research on the unique circadian profiles during hypomania.

The Importance of Learning

Overall, the study’s findings show that the model can predict subsequent emotional episodes by assessing adequate sleep and wakefulness patterns during the patient’s first emotional episode. Since any changes in medication types or dosages can affect patients’ circadian rhythms, the scientists used a new set of experiments to determine the effect of medication changes on the accuracy of the prediction model. in advance. They created a test kit with data from patients who did not change their medication types and dosages after the event began. Using these sets of tests, they confirmed that the accuracy of the prediction model was not related to drug-induced changes in the circadian rhythm.

This study describes the development and validation of a mathematical model that can accurately predict future emotional events solely from binary sleep-wake pattern data. The study also finds that important circadian components, including flow and amplitude shifts, are the main predictors of emotional events, and the slow phase is linked to stressful situations and an advanced stage of depression. associated with manic episodes.

Previous studies conducted at the molecular level have shown a link between circadian rhythms and mood disorders. These studies have shown that changes in the circadian rhythm can lead to abnormalities in serotonergic and dopaminergic circuits by changing the rhythm of the circadian nuclear receptor REV-ERB alpha, which plays an important role in the development of depressive and manic states.

Advantages and Disadvantages

Overall, the predictive model presented in this study opens a new way to diagnose and treat emotional conditions. The main advantage of the model is that it only requires sleep pattern data, which can be collected easily and conveniently with smartphones or wearable devices. Unlike previous studies that relied on simple sleep metrics such as length and capacity, this study included comprehensive sleep parameters and a mathematical model to estimate circadian rhythms.

However, the study has some limitations. It included only patients who followed wearable devices, and the sample was limited to patients with first-episode dementia in South Korea, limiting the generalizability. Furthermore, the nature of the study and reliance on wearable devices, which may be more accurate than laboratory group measurements, were seen as limitations.

The authors suggest that future developments may focus on independent predictive models that correspond to patients’ specific circadian profiles and sleep patterns. This approach can improve the validity and effectiveness of these tools for mental health management.