Universal Digital Health

Lecture 3.0.13 PICOTS Framing & Feasibility

The lecture introduces PICOTS (Population, Intervention, Comparator, Outcome, Timing, Setting) as the essential blueprint for turning vague clinical questions into executable database queries. It stresses that data scientists must translate a clinician’s intuition into strict inclusion and exclusion criteria, using structured electronic health record (EHR) fields rather than free‑text notes. Key insights include defining the population with ICD‑10 codes or lab values, pinpointing the exact data source for interventions (prescription orders, claims, or administration records), selecting appropriate comparators to avoid bias, and choosing measurable outcomes as proxies for clinical improvement. The speaker also outlines feasibility checks—availability, completeness, and format—and warns that even a perfect PICOTS plan can fail if the underlying data are missing or unstructured. A concrete example follows a doctor’s request to evaluate text‑message reminders for diabetic patients. The speaker shows how a broken data bridge (phone numbers vs. medical record numbers) and strict timing windows can cripple the study, and how widening the observation window salvages sample size while trading some precision. The overall implication is that mastering PICOTS and feasibility assessment enables clinicians and data scientists to ask answerable questions, build reliable queries, and generate actionable evidence from real‑world data, accelerating evidence‑based practice.