Models / moph

moph/patient-noshow-predictor

Patient No-Show Predictor

Government-RestrictedLimited riskGov ReuseTabularXGBoostN/A1.3.0

Test in Sandbox

620 5.4K/30d 34

AUROC

0.86

F1 (no-show)

0.71

Precision @ top decile

0.79

Brier score

0.094

About this model

Gradient-boosted classifier that scores the probability an outpatient appointment will be missed, using booking lead time, clinic type, visit history, and reminder-response signals. Designed to drive smarter overbooking and targeted SMS reminders across public health centres.

Intended use

Appointment-scheduling optimisation and reminder prioritisation inside approved MOPH scheduling systems; scores must not be used to deny or deprioritise care.

Training data lineage

Trained on de-identified records from outpatient-appointments-history covering 4.2 million appointments between 2022 and 2025.

Limitations & bias notes

The no-show class is imbalanced (roughly 14% positive) and precision drops for patients with fewer than three prior visits. Behaviour shifts after the 2025 reminder-system upgrade mean scores on pre-2025 workflows are not comparable.

Evaluation metrics

AUROC	0.86
F1 (no-show)	0.71
Precision @ top decile	0.79
Brier score	0.094

#healthcare#xgboost#scheduling#no-show#tabular

Try it

Live sandbox

patient=1184&lead_days=12

fixture

Evaluation only: gated models run against redacted sample data until your entity's entitlement is approved.

Used in workflows

Appointment no-show preventionin-review

Retiring this asset would require these chains to be re-pointed first.

Feedback

Owning entity

moMinistry of Public Health

Updated: 2026-05-21
Latest version: 1.3.0
License: Government Internal Reuse
Access: Gated

Trained on

outpatient-appointments-history →