Predicting ICU Transfers and Other Unforeseen Events (PICTURE)-Pediatric

Pilot Randomized Controlled Trial of PICTURE-Pediatric

The purpose of this study is to evaluate the effectiveness and user satisfaction of the study teams early warning system, called PICTURE, which utilizes artificial intelligence (AI) techniques and algorithms to identify patient deterioration on pediatric units within Mott Children's Hospital.

In this pilot study the patient care team will review the PICTURE information and alerts. Morning rounds will be partially informed by the PICTURE scores and the scores will be included in the hand off notes for the patients with a red score.

The primary purpose of this study is to test the hypothesis that the combination of the PICTURE-Pediatric model, the proposed workflow and the proposed interface results in at least 80% compliance.

No participants will be consented as the Institutional review board has approved a waiver of consent for THE clinicians and the patients information being reviewed.

The enrollment numbers will include only the clinicians.

Study Overview

• Status: Recruiting
• Conditions: Clinical Deterioration
• Intervention / Treatment: Device: PICTURE-Pediatric scores and alerts; Other: Control Arm - Scores and alerts hidden

Detailed Description

PICTURE Pediatric evaluates large amounts of patient data, labs, monitored data, and vital signs (all variables that are measured as part of routine clinical care and that are stored in the electronic health record in real-time); it does this every 15-minutes and delivers a deterioration risk index score that has been shown to outperform Pediatric Early Warning Score (PEWS), and other similar scoring systems in identifying patients at risk for clinical deterioration. Clinical implementation of PICTURE Pediatric could identify patients at high risk of deterioration early enough to intervene and decrease the need for transfer to the Pediatric Intensive Care unit (PICU), and it could also reduce morbidity and hospital length of stay. This in turn could also improve bed availability for patients awaiting admission in the emergency department and improve efficient transfer out of the PICU, all of which improves care and reduces cost for the patients, their families, and the hospital system.

This protocol also does not mandate or require clinical decisions based on alerts generated (e.g. there is no mandate to transfer to the Intensive Care unit, perform an intervention, or change therapy/clinical course based on PICTURE alerts); rather, the clinical team will decide whether additional investigation, therapy, intervention, or transfer is required after alerts are generated based on staff's clinical experience, knowledge and with consultation with an attending physician when applicable. In case of device failure, there would be no deterioration risk scores generated and the clinical teams would simply continue to follow the current workflow.

• Study Type: Interventional
• Enrollment (Estimated): 375
• Phase: Not Applicable

Contacts and Locations

• Study Contact: Name: Jay Semerad; Phone Number: 734-647-4751; Email: jsemerad@umich.edu

Study Locations

• United States
  • Michigan
    • Ann Arbor, Michigan, United States, 48109
      • Recruiting
      • University of Michigan
      • Contact: Jay Semerad; Phone Number: 734-647-4751; Email: jsemerad@umich.edu
      • Principal Investigator: Rodney Daniels, MD, FAAP

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Child; Adult
• Accepts Healthy Volunteers: No

Description

Inclusion Criteria - Patients:

• Age ≥30 days and ≤25 years

• Be on the general care wards of Mott Children's Hospital or boarding in the Emergency Department (ED), but admitted to a pediatric floor medical team. These will be enforced by these criteria:

  1. Patient class: not emergency b. Patient level of care: not Intensive Care unit (ICU) c. Patient locations (per protocol)

     Exclusion Criteria - Patients:

• Patients that do not meet the inclusion criteria

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Supportive Care
• Allocation: Randomized
• Interventional Model: Factorial Assignment
• Masking: None (Open Label)

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: PICTURE-Pediatric scores and alerts; This arm will happen for two four-week periods.	Device: PICTURE-Pediatric scores and alerts; Patients that satisfy the inclusion criteria will receive PICTURE-Pediatric scores and alerts. Scores are numbers that indicate the likelihood that the patient may experience deterioration in the next 24 hours, i.e., the higher the score, the higher the likelihood of deterioration. Alerts will be sent through pages or phone notifications, depending on the user's current method of receiving non-routine communications. Scores are also available through electronic medical record as a column that can be wrenched into the patient lists, in a patient summary report and as a vital and/or a flowsheet. The scores for patients with an alert (scores that have surpassed the corresponding alert thresholds) will be colored accordingly (yellow or red) in the electronic medical record. Various staff on the clinical team will use this data to help determine intervention and or escalation depending on the alerts.
Other: Control Arm; During the control arm, the PICTURE scores and alerts will not be shown. This arm will happen for two four-week periods.	Other: Control Arm - Scores and alerts hidden; The morning rounds and prerounding procedures will revert back to the baseline procedures before the trial. All care teams will be notified about this transition in advance and will be directed to revert back to the standard of care.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Compliance with the model workflow as measured by a pop-up questionnaire after each model alert aggregated as a proportion -clinicians	Responses to the questions measure compliance with the workflow. Compliance will be measured as binary outcomes (i.e., compliant and noncompliant) and aggregated as a proportion.	within 30 minutes after each alert during a patient's admission

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Acceptability of Intervention Measure (AIM)-clinicians	There are 4 questions that the clinicians will answer about intervention with options from completely agree (5) to completely disagree (1). The range of scores are 4-20 with a higher score indicating greater acceptability.	At the end of the study (approximately 4 months after the start of the enrollment)
Intervention Appropriateness Measure (IAM)-clinicians	There are 4 questions that the clinicians will answer about intervention with options from completely agree (5) to completely disagree (1). The range of scores are 4-20 with a higher score indicating greater appropriateness.	At the end of the study (approximately 4 months after the start of the enrollment)
Feasibility of Intervention Measure (FIM)-clinicians	There are 4 questions that the clinicians will answer about intervention with options from completely agree (5) to completely disagree (1). The range of scores are 4-20 with a higher score indicating greater feasibility.	At the end of the study (approximately 4 months after the start of the enrollment)

Collaborators and Investigators

• Sponsor: University of Michigan
• Investigators: Principal Investigator: Rodney Daniels, University of Michigan

Study record dates

Study Major Dates

• Study Start (Actual): February 9, 2026
• Primary Completion (Estimated): January 1, 2027
• Study Completion (Estimated): January 1, 2027

Study Registration Dates

• First Submitted: December 23, 2025
• First Submitted That Met QC Criteria: December 23, 2025
• First Posted (Actual): December 26, 2025

Study Record Updates

• Last Update Posted (Actual): March 5, 2026
• Last Update Submitted That Met QC Criteria: March 3, 2026
• Last Verified: March 1, 2026

More Information

Terms related to this study

• Keywords: Questionnaire; Acceptability; Usability; Artificial Intelligence model; Real-time alerts; Workflow compliance
• Additional Relevant MeSH Terms: Pathologic Processes; Disease Attributes; Disease Progression; Pathological Conditions, Signs and Symptoms; Clinical Deterioration
• Other Study ID Numbers: HUM00274862

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: Yes
• product manufactured in and exported from the U.S.: No