Interindividual differences in digital phenotypes of major depressive disorder: A passive sensing study using smartphone and wearable sensor data.

A two-profile solution demonstrated best fit for digital biomarker data collected from 297 individuals with MDD.

• Participants were 297 individuals diagnosed with major depressive disorder.
• Data were collected from smartphones and Garmin smartwatches using passive sensing.
• Digital biomarkers included sleep patterns, physical activity, screen time, social engagement, and heart rate variability.
• Latent profile analysis was used to identify distinct digital phenotypes.

Profile 1, labeled 'average in every way,' comprised the majority of the MDD sample at 85.7%.

• Profile 1 represented 85.7% of the 297-person sample.
• This profile was characterized by average values across all digital biomarkers.
• Profile 1 served as the reference group for comparisons with Profile 2.

Profile 2, labeled 'deficient sleep, chronically low heart rate variability, and low social engagement,' comprised 14.3% of the sample.

• Profile 2 represented 14.3% of the 297-person sample.
• This profile was distinguished by deficient sleep, chronically low heart rate variability, and low social engagement.
• These features were identified as objective behavioral and physiological markers distinguishing this subgroup.

The two digital phenotype profiles did not significantly differ on MDD symptom severity.

• The estimate for MDD symptom severity difference between profiles was 0.322 (S.E. = 0.767, p = 0.444).
• This non-significant result suggests digital phenotypes do not map directly onto traditional symptom severity measures.
• Self-reported MDD severity was used as the clinical comparator.

Profile 2 had lower social and occupational functioning compared to Profile 1, though this did not survive correction for type I error.

• The estimate for social and occupational functioning difference was -5.309 (S.E. = 2.321, p = 0.023).
• This association was nominally statistically significant but was no longer significant after correcting for type I error.
• The authors noted that sleep dysregulation, low heart rate variability, and low social engagement 'seem to be important indicators of potential social and occupational impairments.'

Previous research on MDD heterogeneity has relied on self-reported symptoms, with few studies using objective passive sensing data to phenotype MDD.

• The authors note that prior studies identified distinct MDD subtypes using self-reported symptoms.
• Passive sensing data from smartphones and wearables can capture 'objective behavioral and physiological patterns, potentially revealing distinct digital phenotypes of MDD.'
• This study was framed as addressing a gap in the use of objective digital biomarkers for MDD phenotyping.

The authors recommend future research incorporate additional digital biomarkers and larger, more diverse samples to validate digital phenotypes of MDD.

• The current sample was limited to 297 individuals, which may limit generalizability.
• The authors called for validation 'against other clinical severity metrics in larger, more diverse samples.'
• Incorporating additional digital biomarkers was identified as a priority for refining MDD digital phenotype identification.