Undiagnosed Pediatric Primary Immunodeficiency Symptom Pattern Recognizer

Post-Joenja approval (first-in-class pediatric APDS treatment, ages 4-11) + 7-year median APDS diagnostic delay. Parent-facing symptom-log app classifies patterns against PIDD signatures and flags for immunologist referral.

Impossibility Negation: 'You cannot recognize undiagnosed primary immunodeficiency from parent-observable symptoms alone.' AI pattern-triage on parent-logged symptom histories has become plausible for the referral-trigger question.

IDF symptom questionnaire and JMF '10 warning signs' are static self-assessment forms. Ada Health and K Health cover generic symptom triage but are not tuned for PIDD signatures. No dedicated pediatric PIDD longitudinal AI symptom tracker found.

Frequency Trap: PIDD symptoms occur monthly at most — subscription-recurring economics on parent side are impossible.

Accuracy Cliff: medical triage on a serious condition. FDA Software as Medical Device framing likely Class II. If app says 'no PIDD' and child has PIDD, delay costs years.

Adoption barrier (structural adoption barrier): parents of chronically-sick children carry medical trauma. Repeated triage without diagnostic finality is often unwelcome. Only a subset of highly engaged parents want ongoing pattern-tracking.

Pivot: physician-facing decision support on EMR encounter data for pediatric practices. Removes parent-side frequency trap and accuracy cliff.

Killed by three-way stack: frequency trap + accuracy cliff + adoption barrier. Physician-side pivot fails on Block 4 template ('AI generates the diagnostic artifact for regulated vertical') and occupied EMR-adjacent market (Bright Health, Notable, Suki, DrChrono).