What the work involves
You will be handed adolescent eating disorder case material — presentations, vitals, weight trajectories, screening scores, family and social context — and asked to produce written clinical evaluations that a model can learn from. In practice that means characterising the presentation against DSM-5, deciding whether the case meets criteria for urgent medical escalation under a framework such as MEED, judging whether a proposed treatment modality (FBT, CBT-E, day patient, inpatient stabilisation) fits the adolescent's clinical stage, and flagging comorbidities and safeguarding concerns that change the risk picture. A second strand of the work is critique: reading an AI-generated or peer-written assessment and identifying where it under-calls bradycardia, over-relies on BMI, misses ARFID or atypical anorexia, or gives advice a non-specialist could act on unsafely.
Writing quality matters as much as clinical accuracy. Your output is read by non-specialists and consumed as training signal, so an evaluation that says "escalate" without stating the physiological threshold that triggered it is close to useless. Expect to justify each judgment against a named guideline or parameter.
What the platform screens for
- Recency of adolescent caseload. Reviewers want people who have managed real adolescent ED patients within roughly the past two years, not historical exposure.
- Instrument fluency under follow-up. EDE-Q, SCOFF, MEED risk parameters — screeners probe whether you know what these tools actually measure and where they fail (SCOFF's poor performance in adolescents and in males, EDE-Q norms, MEED's shift away from BMI-centric thresholds).
- Calibration, not caution. Marking every case high-risk scores no better than marking none. Graders look at whether your escalation threshold matches guidance.
- Written English at C1–C2 level. Evaluations are the deliverable; unclear prose is a fail regardless of clinical strength.
- Licence in good standing and willingness to state jurisdiction and registration details.
Logistics
Fully remote, contractor basis, asynchronous. Tasks are claimed from a queue rather than scheduled, so hours are flexible, but micro1 states it fills roles within 48 hours and expects first tasks within 24–48 hours of onboarding — availability is genuinely part of the assessment. Most contributors work in blocks of a few hours; volume varies with client demand and is not guaranteed. The $45–95/hr band is what has been observed on this listing and typically tracks seniority, specialist training (formal FBT or CBT-E certification), and calibration performance on early tasks. No prior AI experience is expected or required.
