ML-Based Intervention for Pediatrics

Overseen By Agata Wolochacz, BMSc

Age: < 65

Sex: Any

Trial Phase: Academic

Sponsor: The Hospital for Sick Children

Disqualifiers: Prior pharmacogenetic testing, ICU admission, others

No Placebo Group All trial participants will receive the active study treatment (no placebo)

Compensation: Varies

Number of Visits: 1

What You Need to Know Before You Apply

What is the purpose of this trial?

This trial explores whether a machine learning (ML) model can improve medication prescribing for children. It aims to determine if the model can predict when a child might need a specific medication and whether this leads to better use of genetic testing. The goal is to see if this approach results in more children receiving the correct medicine or dose. Children staying at The Hospital for Sick Children who haven't undergone genetic testing or received certain medications before might be suitable candidates. As an unphased trial, this study offers participants a unique opportunity to contribute to innovative research that could enhance pediatric care.

Do I need to stop my current medications for this trial?

The trial information does not specify whether you need to stop taking your current medications. It focuses on predicting and providing pharmacogenetic testing for future medication needs.

What prior data suggests that this ML-based intervention is safe for pediatric use?

Research has shown that using machine learning (ML) in children's healthcare is generally safe. Studies have found that ML can predict health outcomes more accurately without risking patient safety. For example, ML tools can forecast critical health events more effectively than traditional methods, aiding doctors in making informed decisions without interfering with treatment.

Guidelines, such as the ACCEPT-AI framework, ensure the safe use of children's health data in ML research. These guidelines protect patient information and promote ethical practices. While the technology itself does not directly impact health, it supports medical professionals and enhances care, not replace them.

In summary, using ML to predict medication needs or outcomes in children is considered safe, with measures in place to protect data and uphold ethical standards. ^{1 2 3 4 5}

Why are researchers excited about this trial?

Researchers are excited about this ML-based intervention for pediatric care because it uses machine learning to predict which children might benefit from targeted medications shortly after being admitted to a hospital. Unlike traditional treatments that rely heavily on symptomatic observation and physician experience, this approach leverages data-driven insights to make personalized medication decisions. This innovative use of technology could lead to quicker and potentially more effective treatments tailored to individual needs, improving outcomes and reducing the time children spend in the hospital.

What evidence suggests that this ML-based intervention is effective for optimizing drug therapy in pediatrics?

Research shows that machine learning (ML) helps doctors make better treatment decisions in children's healthcare. Studies have found that ML assists with tasks like early detection of health issues and accurate diagnoses, leading to improved health outcomes for kids. For example, ML can identify unusual growth patterns in children, allowing for early and personalized care. In this trial, participants will join an experimental arm where an ML model predicts which children might soon need specific medications. This approach aims to ensure timely genetic tests and medication adjustments. Early findings suggest that ML can streamline this process, potentially leading to safer and more effective treatments for young patients. ^{1 3 5 6 7}

Who Is on the Research Team?

Lillian Sung, MD, PhD

Principal Investigator

The Hospital for Sick Children

Are You a Good Fit for This Trial?

This trial is for pediatric patients who are at high risk of needing a medication that could be influenced by their genetic makeup within the next three months. Specific eligibility criteria have not been provided.

Inclusion Criteria

Inpatient at The Hospital for Sick Children

Exclusion Criteria

Expected hospital discharge is prior to midnight on the day of admission

I have had genetic testing or received targeted medication before.

I am currently admitted to the ICU.

Timeline for a Trial Participant

Screening

Participants are screened for eligibility to participate in the trial

1 day

1 visit (in-person)

Prediction and Notification

Machine learning model predicts patients at high risk of requiring a targeted medication within the next three months. The research team is notified of eligible patients each morning.

3 months

Pharmacogenetic Testing

Pharmacogenetic testing is offered and conducted for participants predicted to receive a targeted medication.

Up to 3 months

Follow-up

Participants are monitored for changes in medication choice or dose based on pharmacogenetic results.

3 months

What Are the Treatments Tested in This Trial?

Interventions

• ML-based intervention

Trial Overview

The study is testing whether a machine learning model can improve drug therapy by predicting which children will need pharmacogenetic testing before they receive certain medications, and if this leads to changes in medication choice or dosage.

How Is the Trial Designed?

Group I: ML model Experimental Treatment 1 Intervention

Participants predicted by an ML model to receive a "targeted medication" within three months following admission.

Who Is Running the Clinical Trial?

The Hospital for Sick Children

Lead Sponsor

Trials

724

Recruited

6,969,000+

Published Research Related to This Trial

A study of 106,122 individual case safety reports (ICSRs) for pediatric patients (0-18 years) submitted to the FDA from 2004 to 2011 revealed that 'nervous system drugs' were the most commonly reported, highlighting the need for focused safety monitoring in this category.

The analysis showed significant differences in reported drug classes and adverse events across different age groups, suggesting that drug safety evaluations for children should be tailored to specific age categories to improve signal detection and safety assessments.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Pediatric Drug Safety Surveillance in FDA-AERS: A Description of Adverse Events from GRiP Project. de Bie, S., Ferrajolo, C., Straus, SM., et al.[2018]

In a study analyzing pediatric hospitalizations, 24 patients experienced 29 adverse events (AEs) related to medical care, with a rate of 3.61% across all pediatric hospitalizations, which is lower than the 6.4% rate seen in non-elderly adults.

A significant 65.5% of these adverse events were deemed preventable, highlighting the need for targeted prevention strategies, especially since medication-related events accounted for 37.9% of the incidents.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

[Clinical safety paediatric patients]. Requena, J., Miralles, JJ., Mollar, J., et al.[2012]

A simulation training program significantly improved the self-efficacy of medical students and interns in identifying and reporting patient safety hazards, indicating its effectiveness as a training tool.

Participants were particularly good at recognizing patient misidentification hazards, but often missed safety issues related to electronic health records (EHRs), highlighting a need for targeted interventions in that area.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

"Good Catch, Kiddo"-Enhancing Patient Safety in the Pediatric Emergency Department Through Simulation. Shaikh, U., Natale, JE., Till, DA., et al.[2023]

In a systematic review of 17 studies involving children, the incidence of adverse drug reactions (ADRs) in hospitalized children was found to be 9.53%, with severe reactions accounting for 12.29%, highlighting the significant risk of ADRs in this population.

The study revealed that 2.09% of pediatric hospital admissions were due to ADRs, with 39.3% of these being life-threatening, indicating a critical need for improved drug safety monitoring and preventive strategies in pediatric healthcare.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Incidence of adverse drug reactions in paediatric in/out-patients: a systematic review and meta-analysis of prospective studies. Impicciatore, P., Choonara, I., Clarkson, A., et al.[2023]

Citations

1.

pmc.ncbi.nlm.nih.gov

pmc.ncbi.nlm.nih.gov/articles/PMC11818243/

Machine Learning in Pediatric Healthcare: Current Trends ...

This narrative review explores the current trends, applications, challenges, and future directions of ML in pediatric healthcare.

2.

publications.aap.org

publications.aap.org/pediatrics/article/156/Supplement%201/e2025070739L/203521/Using-Artificial-Intelligence-and-Machine-Learning

Using Artificial Intelligence and Machine Learning to ...

This is a welcome development that provides a robust, ethical scaffolding to guide the use of children's health care data in AI and ML ...

3.

e-cep.org

e-cep.org/m/journal/view.php?number=20125555822

Artificial intelligence in pediatric healthcare

AI and ML are transforming pediatric subspecialties by augmenting specific clinical tasks such as early detection, precision diagnostics, and ...

4.

researchgate.net

researchgate.net/publication/388470522_Machine_Learning_in_Pediatric_Healthcare_Current_Trends_Challenges_and_Future_Directions

Machine Learning in Pediatric Healthcare: Current Trends ...

Results: ML has demonstrated promise in diagnostic support, prognostic modeling, and therapeutic planning for pediatric patients. Applications ...

5.

mdpi.com

mdpi.com/2227-9067/12/3/317

Advancing Pediatric Growth Assessment with Machine ...

A key strength of ML lies in its ability to detect anomalies in children's growth patterns, allowing for early interventions and personalized treatment plans ...

6.

jamanetwork.com

jamanetwork.com/journals/jamanetworkopen/fullarticle/2834661

Machine Learning to Predict Critical Events in Pediatric Care

Studies have demonstrated that ML algorithms offer significant improvements in predicting critical clinical outcomes compared with conventional tools.

7.

nature.com

nature.com/articles/s41746-023-00898-5

Recommendations for the use of pediatric data in artificial ...

ACCEPT-AI is a framework of recommendations for the safe inclusion of pediatric data in artificial intelligence and machine learning (AI/ML) research.