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.