Health & Wellbeing

Fitbits may find use in predicting depression

Fitbits may find use in predic...
The study tracked almost 300 test subjects who wore a Fitbit Charge 2 for two weeks continuously
The study tracked almost 300 test subjects who wore a Fitbit Charge 2 for two weeks continuously
View 1 Image
The study tracked almost 300 test subjects who wore a Fitbit Charge 2 for two weeks continuously
1/1
The study tracked almost 300 test subjects who wore a Fitbit Charge 2 for two weeks continuously

Within certain populations, such as people with highly stressful jobs, it's important to know if individuals may be developing depression. New research suggests that even if people don't realize they're becoming depressed, a simple Fitbit could warn that they are.

For the study, which was conducted by a team from Singapore's Nanyang Technological University, a total of 290 adults (average age 33) were tasked with wearing a Fitbit Charge 2 activity-tracking device for 14 consecutive days. They were told to wear it at all times, other than when bathing, or recharging its battery.

At the beginning and then again at the end of the two-week period, the participants also completed a questionnaire which is widely used to identify people who are becoming depressed. The results of those questionnaires were then combined with data gathered by the Fitbits, and used to train a machine-learning-based computer program called the Ycogni model.

When that program was subsequently used to analyze the Fitbit data alone, it proved to be about 80 percent accurate at predicting which people were most likely to develop depression, and which ones were least likely to do so.

It was observed that the at-risk individuals had more varied heart rates between 2 am and 4 am, and then again between 4 am and 6 am (as measured by the Fitbits). This falls in line with findings from earlier studies, which suggest that variations in heart rate while sleeping could be a valid physiological indicator of depression.

The Fitbits additionally indicated that the at-risk test subjects tended to have a wider variation in waking times and bedtimes. Again, it has previously been observed that people suffering from depression aren't as good at following daily sleeping and waking routines.

"Our study successfully showed that we could harness sensor data from wearables to aid in detecting the risk of developing depression in individuals," said Prof. Josip Car, who led the study along with Assoc. Prof. Georgios Christopoulos. "By tapping on our machine learning program, as well as the increasing popularity of wearable devices, it could one day be used for timely and unobtrusive depression screening."

The research is described in a paper that was recently published in the journal JMIR mHealth and uHealth. It should be noted that a University of Vermont study previously used wearable motion sensors to detect depression in children.

Source: Nanyang Technological University

3 comments
3 comments
paul314
So did the people identified by either the questionnaire or the algorithm ultimately become depressed?
2Hedz
The author should note: HRV or heart rate variability being higher is a measure of healthy individuals. Not the opposite. What is measured is the difference in successive peaks of an ECG. This is a common metric from wearables, though the accuracy of these devices for such a measurement is still questionable. Direct chest monitors such as the Polar H7 or H10 are considered the gold standard. It is good to have high HRV. It is likely that the study showed lower HRV in more depressed subjects. The data is thin correlating HRV with mental health however. Machine learning definitely will help! And has shown promise in earlier studies.

Differently, having a bigger overall swing in average heart rate in the day vs night, is something known as "rebound". And it is no surprise that it is higher in patients showing signs of depression. This occurs when the sympathetic or parasympathetic nervous system jumps to back to whatever state it is most comfortable with, but overcompensates. Panic attacks may be caused by sympathetic rebound. IBS or migranes can be caused by parasympathetic rebound.

These devices show great promise for mental health monitoring in the future.
2Hedz
I should correct my previous comment. The study did not measure HRV. Instead measured "IS" and "IV", both circadian rhythm metrics, and based on average heart rates. So my previous comment on rebound is applicable but the comment on HRV, although correct, is not applicable to this study.