💊 AI Prescription Interaction Risk Scanner Simulator

Loading the Active Medication List

Every safety scan starts with a complete picture of what a patient already takes.

• 5+: Avg meds per senior patient (polypharmacy threshold)
• ~1.3M: Preventable ADEs yearly (US) (ER visits)
• >14k: Known drug pair interactions (in reference databases)
• ~40%: Reconciliation error rate (at care transitions)

Why the full list matters

One missing drug can hide a dangerous pair.

Sources feeding the list

Pharmacy claims, EHR orders, and patient-reported meds merge.

Normalizing drug identities

Brand names map to generic RxNorm codes for matching.

A New Prescription Enters the Picture

Adding one drug means testing it against every drug already present.

• 1–2: Avg new Rx per visit (per prescriber encounter)
• ~90%: Alert fatigue override rate (of low-value alerts ignored)
• ~12: High-risk drug classes (flagged for extra scrutiny)
• <1 sec: Time to scan one drug (automated check)

The trigger event

A prescriber or pharmacist submits a new order.

Candidate drug profile

Class, metabolic pathway, and known contraindications load.

Why timing matters

Catching risk before dispensing prevents harm.

Checking the New Drug Against Every Existing One

The AI runs one comparison per existing medication, exhaustively.

• 10: Checks for a 10-drug list (one per existing drug)
• ~5 ms: Interaction lookup latency (per pair, cached rules)
• 4: Mechanism categories checked (metabolic, additive, receptor, renal)
• ~8%: False positive rate (tuned) (industry benchmark)

Exhaustive pairing

Every existing drug gets its own comparison pass.

Rule and model hybrid

Curated rules plus a learned risk model score each pair.

Mechanism awareness

Enzyme competition, additive effects, and renal load all count.

Severity Scoring for Every Flagged Pair

Not all interactions are equal — each gets a graded severity score.

• 4: Severity tiers used (minor to contraindicated)
• 100%: Contraindicated pairs blocked (hard stop required)
• ~60%: Moderate pairs need review (require pharmacist sign-off)
• 6+: Scoring factors weighed (dose, renal, age, mechanism)

Severity ladder

Minor, moderate, major, and contraindicated tiers apply.

Patient-specific weighting

Age, renal function, and dose shift the score.

Actionable output

Each score maps to a recommended clinical action.

Scan Complete — Cleared or Flagged

The list ends either fully cleared or annotated with specific risks.

• ~70%: Scans returning zero flags (typical outpatient orders)
• ~2%: Contraindicated hits blocked (of all new orders)
• ~4 min: Clinician review time saved (per order, automated)
• always: Repeat scan on any edit (list changes retrigger scan)

Clear outcome

No flags means the order proceeds automatically.

Flagged outcome

Each flagged pair routes to pharmacist or prescriber review.

Continuous rescanning

Any future list change reruns the entire scan.