Tools · Decision support

Differential Diagnosis Engine

The DDx engine from Module 7 in a clinic-mode wrapper. Toggle the clinical features you have observed and the candidate diagnoses re-rank live, each with the features supporting and opposing it. The intent is interpretability — every ranking is explicable from the visible features, not a black-box output.

Clinical use note. This tool is a teaching scaffold and a structured prompt for the differential — not a diagnostic replacement for clinical judgement. Weights are ordinal expert priors (+3 criterion-level, +1 supportive, −2 actively excluding), not probabilities. Two diagnoses with the same fit percentage are not equally likely — they are equally fitting their respective feature profiles. The clinician's job is to interrogate the ranking, not accept it.

Interactive Differential Engine

Toggle the clinical features you have observed in the patient. The candidate diagnoses re-rank live, each with the features supporting and opposing it. Designed as a teaching scaffold — the engine does not replace clinical reasoning, but it makes the differential structure of cervicogenic dizziness visible.

No features selected

History

Trigger / pattern

Duration of episodes

Vestibular workup findings

Imaging findings

Associated symptoms

Cervical examination

Select clinical features on the left to see ranked differentials.
Figure 7.1 — Interactive differential-diagnosis engine. Each diagnosis carries a feature profile drawn from current consensus criteria (Lempert 2022 for vestibular migraine, Staab 2017 for PPPD, López-Escámez 2015 for Ménière's, Strupp 2022 for vestibular neuritis, von Brevern 2015 for BPPV, Seemungal 2022 for cervical dizziness). Features and weights are visible in lib/ddx.ts; the engine is a transparent scaffold, not a black-box classifier.