Case-retrieval engine
A national case-retrieval engine for acute poisoning management
As a new case is entered, ToxMatch compares it against the national registry and retrieves the most comparable historical cases — each shown with its full management trajectory and final outcome. Decisions are grounded in real, local precedent rather than generic international references.
The problem
A frequent, time-critical emergency
Acute poisoning is a common presentation in Malaysian emergency departments — and one where the first decisions strongly influence the outcome. The treating clinician often has minutes, not hours.
The gap
Data that never reaches the next patient
Case entered→Stored in registry→never reused
There is no rapid way to ask the most useful question: what happened to patients like this one — and how were they managed?
How it works
A new case finds its precedents
STEP 01
Enter the case
Key clinical features are entered as the patient is assessed — agent, vitals, labs, route.
STEP 02
Score similarity
Each registry case is scored against the entry using a weighted clinical feature vector.
STEP 03
Retrieve precedents
The top matching cases are surfaced — ranked by clinical similarity, not recency.
STEP 04
Review trajectories
Each result shows the full management course and outcome of that real case.
247
Registry cases
Drawn from the national pesticide-poisoning registry — paraquat-dense, expanding to other agents.
15+
Clinical features
Weighted similarity across agent, vitals, labs, route, and time-to-contact.
100%
Deterministic
No machine learning, no black-box predictions. Every match is explainable from first principles.
Matching logic
The clinical feature vector
Similarity is computed over explicit features — every match shows why it was selected. No opaque predictive model is involved.
Agent / toxidrome
Route of exposure
Form of poison
Amount ingested
Hours to contact
Age
GCS
Systolic BP
Heart rate
SpO₂
Respiratory rate
Creatinine
Lactate
pH
Urine paraquat
Roadmap
From pesticides to the full toxicology workflow
LIVE TODAY
Pesticide poisoning
NEXT
Full registry integration
NEXT
Medication overdose cohort
NEXT
Multi-centre data feed
National Poison Centre · Universiti Sains Malaysia · Project lead: Ts. Dr. Ahmad Shalihin Mohd Samin
Displays similar historical cases for reference. Not a diagnostic or treatment recommendation. Clinician judgement applies.