Example output
# IRB Protocol Summary
## Title: Digital Phenotyping of Adolescent Mood Variability via Passive Smartphone Sensing
## Principal Investigator: I. Papadakis, PhD (Department of Psychiatry)
---
## 1. Background and Scientific Rationale
Adolescent depression is frequently diagnosed late because episodic clinical encounters capture only a narrow window of mood variability. Passive digital phenotyping — using smartphone sensor and interaction data as proxy measures of behavior — has shown promise in adult populations for detecting subclinical mood deterioration up to two weeks before clinical escalation. Validation in adolescents (ages 14-17) is limited, and the ethical profile is distinct from adult research. This study will assess the feasibility, signal quality, and acceptability of passive digital phenotyping in an adolescent community sample.
## 2. Objectives and Hypotheses
- **Primary:** Determine whether passive sensing features (screen unlock frequency, keystroke dynamics, GPS variance, sleep regularity) correlate with weekly PHQ-A self-report scores over 12 weeks.
- **Secondary:** Characterize participant acceptability and perceived privacy cost.
## 3. Participant Population
- **N:** 120 adolescents aged 14-17, community-recruited.
- **Inclusion:** Owns a personal smartphone used daily; fluent English reader; parent/guardian co-enrolled.
- **Exclusion:** Current psychiatric hospitalization; active safety plan for self-harm at enrollment; shared-device situations where the participant does not control the phone.
- **Recruitment:** School-district partnership and community health clinics. Flyers screened by IRB.
## 4. Procedure
At enrollment, participants install a study app that passively logs sensor and usage data. Weekly, participants complete a PHQ-A (9 items, ~3 min) via in-app prompt. Every 4 weeks, a brief acceptability survey (~8 min). Total active time commitment: ~45 minutes over 12 weeks. Passive logging is continuous but invisible.
## 5. Risks
| Risk | Category | Severity | Likelihood | Mitigation |
|------|----------|----------|------------|-----------|
| Privacy breach of sensor data | Social/Legal | High | Low | End-to-end encryption; on-device aggregation; no raw GPS coordinates transmitted |
| Psychological distress from repeated mood self-report | Psychological | Moderate | Moderate | Validated brief measure; in-app crisis resources on every submission; clinical backstop |
| Identification of suicidality without response | Psychological | Severe | Low | PHQ-A item 9 triggers real-time clinician review protocol within 4 hours |
| Parental coercion to enroll | Social | Moderate | Moderate | Assent administered privately; participant can withdraw without parental notification |
| Incidental discovery of illegal behavior via location data | Legal | Moderate | Low | GPS processed into abstract mobility features only; raw traces never accessed by researchers |
| Stigma if peers learn of participation | Social | Low | Low | No study branding on app; generic icon |
## 6. Benefits
- **Direct to participant:** None guaranteed. Weekly self-monitoring may promote emotional self-awareness in some participants. Participants receive a personal mood-trend summary at study end.
- **Societal:** Evidence base for ethical adolescent digital phenotyping; potential future tools for earlier identification of mood episodes.
## 7. Risk-Benefit Analysis
The study poses more than minimal risk, primarily due to the passive sensing component and the adolescent population. However, the risk is non-trivial specifically because the research is worth doing — weaker studies would yield weaker evidence and prolong the use of poorly-understood digital mental health tools in young people. Mitigations (on-device aggregation, real-time suicidality response, adolescent-private withdrawal) meaningfully reduce the residual risk. The societal benefit of establishing ethical and empirical baselines for adolescent digital phenotyping justifies proceeding, provided the monitoring board reviews every 4 weeks and retains authority to pause.
## 8. Consent Process
Parental permission obtained in writing. Adolescent assent obtained separately, with the researcher explicitly confirming the participant understands they can withdraw at any time without parental knowledge. A one-page plain-language summary is provided. Participants demonstrate understanding via four teach-back questions before consent is recorded.
## 9. Data Protection
All sensor data aggregated on-device into features; raw streams never leave the phone. Features transmitted over TLS to institutional secure server. Data retained for 7 years post-publication per institutional policy, then destroyed. Only the PI and two named analysts have access.
## 10. Compensation
$15 per completed monthly check-in ($60 total), calibrated to avoid undue inducement while respecting participant time. No compensation for passive data alone, to avoid incentivizing device behavior changes.