Methodology
Adaptive screening
Psyche Prescreening uses a construct-adaptive approach. Instead of administering fixed-length instruments from start to finish, we draw from a bank of validated items tagged by psychological construct. Each item you answer updates our estimate across multiple constructs simultaneously, so one response can provide information about depression, sleep, and anxiety at the same time.
Gate-then-drill
Every construct starts with a small number of gate items. If those responses suggest the area is not relevant to you, we move on (costing only 1 to 3 items). If the gate opens, we administer additional items to get a more precise estimate.
Bayesian estimation
We use Bayesian posterior estimation to maintain a probability distribution over each construct as you respond. This lets us calculate confidence intervals and stop administering items for a construct once we have enough certainty, keeping the total screening time to 10 to 20 minutes across dozens of constructs.
Source instruments
Items are drawn from published, validated screening instruments including the PHQ-9, GAD-7, PC-PTSD-5, ASRS-v1.1, AQ-28, AUDIT-C, SCOFF, OCI-R, ISI, and others. Clinical thresholds are based on the published cutoff scores from each instrument's validation studies.
Report generation
Your screening report is generated by an AI model that synthesizes your construct-level scores into a narrative. The model is instructed to never use diagnostic language, never recommend medications, and always include relevant crisis and support resources.
Limitations
Screening instruments are not diagnostic tools. They have false positives (flagging something that is not clinically significant) and false negatives (missing something that is). Adaptive administration can introduce additional measurement error compared to full-length instrument administration. Your results should always be discussed with a qualified professional.