PREDICT-DL: Deep Learning Paediatric ED Admission Prediction Study, Western Australia, Target 35000

Trial Review

The safety and scientific validity of this study is the responsibility of the study sponsor and investigators. Listing a study does not mean it has been endorsed by the ANZCTR. Before participating in a study, talk to your health care provider and refer to this information for consumers Trial registered on ANZCTR

Registration number

ACTRN12626000513314 Ethics application status

Approved Date submitted

11/03/2026 Date registered

24/04/2026 Date last updated

24/04/2026 Date data sharing statement initially provided

24/04/2026 Type of registration

Prospectively registered
Titles & IDs Public title PREDICT-DL: Performance of Paediatric Real-time Emergency Department admission and Inpatient Care Time prediction with Deep Learning Query! Scientific title PREDICT-DL: Performance of Paediatric Real-time Emergency Department admission and Inpatient Care Time prediction with Deep Learning Query! Additional trial identifiers [1] 316761 0 None Query! Universal Trial Number (UTN) Query! Trial acronym Query! Related trial records Query!
Health condition Health condition(s) studied: Emergency Medicine 341061 0 Query! Patient flow 341062 0 Query! Condition category Condition code Public Health 337186 337186 0 0 Query! Health service research Query!

Emergency medicine 337560 337560 0 0 Query! Other emergency care Query!

Intervention/exposure Study type Interventional Query! Description of the intervention or exposure Patient flow prediction modelling with machine learning using triage data. Machine learning will use a ensemble model of XGboost, multilayer perceptron and TabNET with a logistic regression blend and natural language processing to interpret and embed the triage free text.
Outcomes of patient flow predicted are admission from ED, with short stay vs medical vs surgical admissions to also be predicted, as well as the length of stay.
The study will use the previous 5 years of data for training and simulation prior to the intervention, with the intervention then using 6 months of live prospective data. During this time, the bed managers who receive the model outputs will be involved in qualitative interviews as an additional outcome and source of data.
Bed manages will receive hourly predictions (with confidence intervals) on a digital bed occupancy dashboard and they will receive training in how these predictions are created and what they mean via educational sessions and a reference guide available at the time of use. This dashboard is already in clinical use, it just does not provide any ahead of time predictions. It is intended that this can help inform their capacity planning, though resource allocation ahead of the time of actual admission. This will given them information from the time of a patients triage, giving them valuable lead time, but the system will not suggest how to change capacity planning, it will only provide predictions of what the bed demand will be in the upcoming hours. There is no quantitative method of assessing adherence or click through based on system limitations, though assessment will occur via qualitative interviews both early and late in the intervention. Query! Intervention code [1] 333447 0 Early detection / Screening Query! Description of the comparator or control No control group. Query! Control group Uncontrolled Query!
Outcomes Primary outcome [1] 344699 0 Model performance in predicting the required number of beds (admissions) as a composite of both discrimination and calibration metrics below. Query! Timepoint [1] 344699 0 Hourly measurements and assessed over a 6 month intervention period. Query! Primary outcome [2] 345148 0 Prediction of length of stay (LOS - AKA care time) accuracy Query! Timepoint [2] 345148 0 Hourly across the study the value will be recorded and stored, then the rolling and absolute mean will be reported. Query! Secondary outcome [1] 458938 0 Qualitative assessment of tool usefulness Query! Timepoint [1] 458938 0 One interview with each consenting staff member between the 1 to 2 month mark (range due to multiple interviews) and another interview with each staff member at the conclusion of the study, with all exit interviews to occur within two months of the conclusion of the intervention. Query!
Eligibility Key inclusion criteria All presentations to the hospital. Query! Minimum age 0 Years Query! Query! Maximum age 20 Years Query! Query! Sex Both males and females Query! Can healthy volunteers participate? No Query! Key exclusion criteria Nil. Query!
Study design Purpose of the study Prevention Query! Allocation to intervention Non-randomised trial Query! Procedure for enrolling a subject and allocating the treatment (allocation concealment
procedures) Query! Methods used to generate the sequence in which subjects will be randomised (sequence
generation) Query! Masking / blinding Open (masking not used) Query! Who is / are masked / blinded?

Query! Query! Query! Query! Intervention assignment Single group Query! Other design features Query! Phase Not Applicable Query! Type of endpoint/s Query! Statistical methods / analysis Time series of performance metrics, SHAP feature importance, covariate drift, concept drift analysis and qualitative analysis. Query!
Recruitment Recruitment status Not yet recruiting Query! Date of first participant enrolment Anticipated 1/06/2026 Query! Actual Query! Date of last participant enrolment Anticipated 30/11/2026 Query! Actual Query! Date of last data collection Anticipated Query! Actual Query! Sample size Target 35000 Query! Accrual to date Query! Final Query! Recruitment in Australia Recruitment state(s) WA Query!
Funding & Sponsors Funding source category [1] 321358 0 University Query! Name [1] 321358 0 University of Notre Dame, Australia Query! Country [1] 321358 0 Australia Query! Primary sponsor type Individual Query! Name Dr Ethan Williams, PhD Candidate at the University of Notre Dame Australia and Medical Doctor at the Child and Adolescent Health Service, WA Query! Address Query! Country Australia Query! Secondary sponsor category [1] 324098 0 University Query! Name [1] 324098 0 University of Notre Dame, Australia Query! Address [1] 324098 0 Query! Country [1] 324098 0 Australia Query! Secondary sponsor category [2] 324330 0 Charities/Societies/Foundations Query! Name [2] 324330 0 Stan Perron Charitable Foundation Query! Address [2] 324330 0 Query! Country [2] 324330 0 Australia Query!
Ethics approval Ethics application status Approved Query! Ethics committee name [1] 319892 0 Child and Adolescent Health Service Human Research Ethics Committee Query! Ethics committee address [1] 319892 0 https://cahs.health.wa.gov.au/Research/For-researchers/Ethics-and-governance-approval Query! Ethics committee country [1] 319892 0 Australia Query! Date submitted for ethics approval [1] 319892 0 20/10/2025 Query! Approval date [1] 319892 0 10/03/2026 Query! Ethics approval number [1] 319892 0 Query!
Summary Brief summary This is a prospective, real-world proof of concept study evaluating the performance of a continuously deployed machine learning (ML) forecasting system that generates hourly, ED-cohort–level admission and length of stay (LOS) prediction. It is set within a Paediatric Hospital, with data from the Emergency Department (ED) triage comprising the model’s features. The model will predict expected inpatient ward admissions (including surgical vs medical if possible), expected ED Short Stay Unit (ESSU) admissions, and the expected mean inpatient LOS among those expected to be admitted to the ward. The system operates in parallel to usual operations, without significant system demands and is not planned to alter clinical decision-making during the evaluation period. The outcomes for this modelling are intended for the patient flow team and bed managers only, with outputs only occurring on an ED cohort level, not on an individual patient level. Query! Trial website Query! Trial related presentations / publications Query! Public notes Query!
Contacts Principal investigator Name 148750 0 Dr Ethan Williams Query! Address 148750 0 Notre Dame Fremantle, School of Medicine, 32 Mouat St, Fremantle WA 6160 Query! Country 148750 0 Australia Query! Phone 148750 0 +610434032522 Query! Fax 148750 0 Query! Email 148750 0 [email protected] Query! Contact person for public queries Name 148751 0 Ethan Williams Query! Address 148751 0 Notre Dame Fremantle, School of Medicine, 32 Mouat St, Fremantle WA 6160 Query! Country 148751 0 Australia Query! Phone 148751 0 +610864562222 Query! Fax 148751 0 Query! Email 148751 0 [email protected] Query! Contact person for scientific queries Name 148752 0 Ethan Williams Query! Address 148752 0 Notre Dame Fremantle, School of Medicine, 32 Mouat St, Fremantle WA 6160 Query! Country 148752 0 Australia Query! Phone 148752 0 +610864562222 Query! Fax 148752 0 Query! Email 148752 0 [email protected] Query!
Data sharing statement Will the study consider sharing individual participant data? No
No IPD sharing reason/comment: Patient data is not to be stored beyond the study as it will continue to be stored by the hospitals records team. Access to the patient data is only available on request to the ethics committee and relevant data custodians and this is not up to the discretion of the research team. The research team is happy to advise on the specifics of data requests for any other research teams.

What supporting documents are/will be available?

No Supporting Document Provided

Results publications and other study-related documents

Documents added manually

No documents have been uploaded by study researchers.

Documents added automatically

No additional documents have been identified.