For people living with depression, symptoms often change over time. At a doctor’s appointment, it can be hard to remember how you felt months ago, and because these symptoms go up and down, a single appointment may not show the full picture. But an accurate diagnosis is key to getting the best treatment.

One way to collect clearer symptom information could be via smartphones, especially as many people carry their phones with them everywhere. Could details like time spent on certain apps or the number of phone calls help measure depression severity?

In this edition of Tap into Research, we look at a study investigating what data from smartphones might reveal about how severe depression symptoms are.

How were smartphones used in the study?

For two weeks, participants shared data in two ways:

• Actively collected data: People filled in daily surveys on their phones. This included rating their stress levels three times a day, sleep quality in the morning, and providing information on social interactions and nutrition at night.
• Passively collected data: Phones collected information without participants having to input any information themselves, such as:
  • Number of times phone locked and unlocked
  • Number and length of calls
  • Daily time spent on certain apps
  • GPS location, number of places visited, and distance travelled.

What did the study find?

This study uses a statistic called “R-squared” to show how useful certain information is for estimating depression severity. It goes from 0% (not useful) to 100% (perfect). For example, knowing someone’s height might tell you a bit about their weight (maybe 60%), but not all of it. The other 40% could be explained by things like age, sex, and diet. This study used R-squared to see how well survey answers and smartphone data predicted depression severity.

People with more severe depression reported feeling more negative and having a lower quality of social interactions. Daily surveys explained about 35% of how severe depression was.

Using the information collected from people’s smartphones passively, patterns like longer calls and more screen time were associated with higher depression levels. This passive data explained about 20% of depression severity.

Using all the information collected from the daily surveys and passively from people’s smartphones explained about 45% of how severe depression was.

Why does this matter?

This study shows that smartphone data could help assess depression more accurately, but more research is needed with different groups of people, particularly in healthcare settings, to build confidence in these findings. It also raises important questions about how comfortable people feel sharing this smartphone data, showing that working with patients and the public is essential for moving forward in this research.

Read the paper: Investigating smartphone-based sensing features for depression severity prediction: observation study

Related