Related measurement records

Selected delirium risk-prediction models

These models estimate the chance of delirium developing in a defined population. They do not detect current delirium and are kept separate from bedside screening, diagnostic and severity instruments.

A selected collection, not a recommendation

The collection is deliberately labelled selected rather than complete. Inclusion records a named model with reproducible inputs, peer-reviewed development and a defined target population, plus temporal or external validation or a public implementation where identified. It does not show that a model is suitable for a particular service, electronic record or patient.

PRE-DELIRIC delirium prediction model

Related context

A ten-predictor model estimating the probability of developing delirium during an ICU stay. It supports prevention planning and does not detect delirium already present.

Open profile

AWOL

2013

AWOL delirium prediction rule

Related context

AWOL is a four-variable admission risk model for estimating which medical inpatients may develop delirium. It supports prevention planning; it does not detect current delirium.

Open profile

Early PRE-DELIRIC delirium prediction model

Related context

A nine-predictor admission-time model estimating delirium risk across an ICU stay. It is an earlier related model, not a bedside test for current delirium.

Open profile

Automated Delirium Risk Assessment System

Related context

An eleven-predictor EHR system that displays ICU delirium risk categories and was followed after implementation. It predicts future risk and does not replace direct delirium assessment.

Open profile

BDS

2018

Buffalo Delirium Scale

Related context

A home-hospice monitoring measure for changes that may precede delirium. It describes prodromal symptom frequency and does not establish a current DSM delirium diagnosis.

Open profile

DYNAMIC-ICU delirium prediction rule

Related context

A seven-predictor dynamic ICU rule for estimating the risk of developing delirium during admission. It is a prevention and monitoring aid, not a current-delirium screen.

Open profile

AWOL-S

2020

Age, WORLD backwards, Orientation, iLlness severity and Surgery-specific risk

Related context

A perioperative extension of AWOL that estimates postoperative delirium probability using age, brief cognition, illness severity and surgical risk. It predicts risk rather than detecting delirium.

Open profile

MDP

2021

Mayo Delirium Prediction tool

Related context

An automated 19-variable hospital risk model that estimates the probability of developing delirium. It supports prevention and systematic assessment; it does not detect delirium already present.

Open profile

DELIKT

2023

DELIrium risK Tool

Related context

A single-centre EHR-derived score for incident delirium risk in older non-ICU inpatients. It is a prediction model requiring external and implementation validation, not a current-delirium screen.

Open profile

PIPRA

2023

Pre-Interventional Preventive Risk Assessment

Related context

An international preoperative model estimating postoperative delirium risk before selected surgery. It is a risk calculation, not a test for delirium, and commercial relationships require transparent disclosure.

Open profile

Before considering implementation

  • Check that the development and validation populations match the intended setting and time point.
  • Confirm that every predictor can be obtained consistently without information leakage or unavailable future data.
  • Review calibration, discrimination, missing-data handling, external validation and clinical utility—not only a headline area under the curve.
  • Assess local workflow, equity, information governance, maintenance and monitoring before any software or EHR integration.
  • Continue direct clinical assessment when delirium is suspected; a risk estimate cannot rule current delirium in or out.

The catalogue does not attempt to include every one-off regression, nomogram or machine-learning experiment. Submit a named model and its primary source through Contribute and corrections.