Resources

One row per day: total, ambulance and walk-in arrivals, the planted flags, and the planted expected value. From Emergency department demand forecasting.

Every forecast from 24 monthly origins, with 80% intervals for the 2025 origins and the actuals. From Emergency department demand forecasting.

19,998 discharges for whole patients chosen at random. ward_code is the planted leak; leave it out of any model. From 30-day readmission risk model.

Test-period metrics in long format: one row per model, group and metric, including call-list capture and PPV. From 30-day readmission risk model.

The star schema as CSV. Unzip into the folder the CsvFolder parameter points at. From Referral-to-treatment waiting list semantic model.

One row per pathway: 58,050 rows, with clock start, clock stop and stop type. From Referral-to-treatment waiting list semantic model.

Sunday × specialty × priority × wait band: 19,884 rows. From Referral-to-treatment waiting list semantic model.

13,030 rows: every extract row for one appointment in twelve, re-sent copies included. From Outpatient analytics pipeline.

Seven marts as CSV. Rates come with their numerators and denominators. From Outpatient analytics pipeline.

78 synthetic weeks: the value, its moving range, the baseline limits, and one True/False column per special-cause rule, exactly as special_causes returns them. From SPC toolkit, a small Python package.

45 synthetic clinics: counts, rate, planted true rate, z-score, and 99.8% limits and flags with no adjustment, multiplicative and additive. From SPC toolkit, a small Python package.

The star schema as CSV, including the security mapping. Unzip into the folder the CsvFolder parameter points at. From Theatre utilisation semantic model.

One row per list: 6,844 rows with planned and actual times and the minutes lost. From Theatre utilisation semantic model.

One row per booked case: 35,014 rows with times, touch minutes and cancellation reasons. From Theatre utilisation semantic model.

One row per scenario and measure: mean with 95% interval, 10th, 50th and 90th percentiles across 200 simulated years, and the paired change from current practice. From Bed occupancy simulation.

Replication 1 of all six scenarios, hour by hour for 52 weeks: patients in a bed and patients waiting for one. From Bed occupancy simulation.

1,287 synthetic letters from 11 practices: labels, planted-difficulty flags, keyword and model predictions, probabilities and the routing decision. From Referral triage text classifier.

Every appointment for a random 40% of trial patients (7,887 rows): arm, text delivery status, outcome, and the planted DNA probability under each arm. From Appointment reminder experiment.

Every estimate on this page with its 95% interval, p-value and planted value. From Appointment reminder experiment.

All 18 measures with their descriptions. Paste into DAX query view and run. From Referral-to-treatment waiting list semantic model.

All 17 measures, the 6 calculation items as comments and two test queries. Paste into DAX query view. From Theatre utilisation semantic model.

From Emergency department demand forecasting.

From 30-day readmission risk model.

From Referral-to-treatment waiting list semantic model.

From Outpatient analytics pipeline.

From Content gate: LLM review with code-owned verdicts.

From SPC toolkit, a small Python package.

From Theatre utilisation semantic model.

From Bed occupancy simulation.

From Referral triage text classifier.

From Appointment reminder experiment.

DNA rate, cancellation rates, slot utilisation and new-to-follow-up ratio. From Outpatient analytics pipeline.

Column tests and reconciliations in the shape of a dbt properties file. From Outpatient analytics pipeline.