Bayesian (log-odds) framework for estimating diagnostic yield of blood-derived exome or genome sequencing in pediatric neurodevelopmental presentations.
Bayesian framework. Each feature contributes a log-odds shift toward the hypothesis "this patient has a diagnostically-identifiable Mendelian etiology by blood-derived exome." Combination respects phenotype overlap — gene-space shared between phenotypes means features within the same group don't stack linearly. Calibrated to Srivastava 2019 meta- analysis and subtype-specific yields from the brain malformation literature (see Evidence anchors below).
Testing modality
The testing approach materially affects yield. Trio over proband adds de novo identification (~7-10% in standard NDD cohorts). Adding CMA to trio exome catches pathogenic CNVs that exome can miss (~5-7%). WGS adds only modest incremental yield over WES + CMA (~2-6%, and not statistically significant in several within-cohort studies).
Standard contemporary approach. ~7-10% absolute yield lift over proband-only by enabling de novo identification and inheritance phasing. Soft ceiling ~63% (still limited by exome coverage gaps and CNV detection).
Additional analyses (optional add-ons)
Optional add-ons that incrementally raise yield beyond the standard sequencing modality.
Core phenotype
Recognizable syndrome patterns or facial features pointing to defined gene groups. Replaces older 'dysmorphism' terminology.
Loss of acquired milestones (distinct from developmental plateau). Points to specific gene groups: MECP2, CDKL5, lysosomal, leukodystrophy, mitochondrial.
Associated features
The three epilepsy options are mutually exclusive — select the highest applicable. Pick DEE/drug-resistant for intractable seizures; neonatal/early-infantile for seizures within the first weeks of life; general epilepsy for everything else.
Brain malformations
Subtype-specific yields. If multiple subtypes are present (rare but possible), the strongest contributes fully and others at reduced weight. Note tissue-source caveats for FCD and HME.
Family history
Estimated yield
Trio exome (proband + both parents)
~31%· modeled ~28–34%
Near baselineConfidence ●●●○ well-anchored
Anchored to direct cohort evidence.
Modeling range, not a statistical confidence interval.
Counseling framing: Yield near the average for NDD exome (Srivastava 2019: 31% isolated, 36% overall). Reasonable to proceed; expect roughly 1 in 3 chance of definitive diagnosis. Trio testing preferred when accessible.
Estimate for resident teaching / family counseling — not validated for individual prediction (PredWES, van der Sanden 2022). Full disclaimer in About tab below.