Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Mortality risk prediction from dense PICU data from patients with suspected infection: data-derived physiological

Tom Velez1, Oluwakemi Badaki-Makun2, Danielle Hirsch3

  • 1Computer Technology Associates, Cardiff, CA, USA.

International Journal of Infectious Diseases : IJID : Official Publication of the International Society for Infectious Diseases
|June 18, 2026
PubMed

Related Concept Videos

Steps in Outbreak Investigation01:18

Steps in Outbreak Investigation

In the ever-evolving field of public health, statistical analysis serves as a cornerstone for understanding and managing disease outbreaks. By leveraging various statistical tools, health professionals can predict potential outbreaks, analyze ongoing situations, and devise effective responses to mitigate impact. For that to happen, there are a few possible stages of the analysis:
Kaplan-Meier Approach01:24

Kaplan-Meier Approach

The Kaplan-Meier estimator is a non-parametric method used to estimate the survival function from time-to-event data. In medical research, it is frequently employed to measure the proportion of patients surviving for a certain period after treatment. This estimator is fundamental in analyzing time-to-event data, making it indispensable in clinical trials, epidemiological studies, and reliability engineering. By estimating survival probabilities, researchers can evaluate treatment effectiveness,...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

How Correct is AI for Infant Safe Sleep Advice? Evaluating Accuracy of ChatGPT, Gemini, and Claude Against AAP Guidelines.

Academic pediatrics·2026
Same author

Cytokine-based strategies to improve prognostic enrichment of pediatric ARDS.

Critical care (London, England)·2026
Same author

Impact of breast reconstruction on sexual function and body image in breast cancer survivors: A systematic review.

JPRAS open·2026
Same author

PCT-Anchored Machine Learning for Pre-Culture Identification of Gram-Negative Sepsis in Children: A Four-Site Study.

Shock (Augusta, Ga.)·2026
Same author

Perspectives on artificial intelligence supported firearm access screening and storage counseling in pediatric emergency care.

The American journal of emergency medicine·2026
Same author

Assessing the capability of large language models in answering pediatric critical care board-style questions.

Scientific reports·2026
Summary

Machine learning models using routine pediatric intensive care unit (PICU) data can predict early mortality risk in critically ill children. These models outperform traditional severity scores, highlighting the power of physiologic dynamics and treatment data.

Area of Science:

  • Critical Care Medicine
  • Pediatric Health
  • Machine Learning in Healthcare

Background:

  • Early identification of mortality risk in critically ill children with suspected infections is crucial but challenging.
  • The incremental value of established severity scores (Systemic Inflammatory Response Syndrome [SIRS], pediatric Sequential Organ Failure Assessment [pSOFA], Phoenix) over data-derived features in machine learning models is unclear.

Purpose of the Study:

  • To compare the predictive value of physiologic dynamics, treatment features, and continuous laboratory trends against expert-derived severity representations for 24-hour pediatric intensive care unit (PICU) mortality.
  • To develop a parsimonious and interpretable machine learning model for early mortality prediction.

Main Methods:

  • Analysis of 3,310 PICU encounters using first-24-hour data, with 107 deaths (3.2% mortality).
Keywords:
InfectionPediatric intensive care unitPhysiologic dataTemporal resolution

Related Experiment Videos

  • Development of sequential XGBoost models using a staged, domain-structured feature selection strategy and 5-fold cross-validation.
  • Performance evaluation using Area Under the Receiver Operating Characteristic Curve (AUROC), Area Under the Precision-Recall Curve (AUPRC), Brier score, and specificity at 80% sensitivity, with SHAP-guided pruning.

    • Main Results:

    • Vital sign extrema provided initial discrimination; incorporating variability and slope features enhanced calibration and specificity.
    • Features related to vasopressor and supplemental oxygen timing/intensity significantly improved AUROC and AUPRC.
    • Continuous laboratory data and temporal trends further boosted predictive performance, outperforming expert-derived severity scores (SIRS, pSOFA, Phoenix).

    Conclusions:

    • Routinely collected hourly PICU data from the initial 24 hours are sufficient for XGBoost models to identify deterioration patterns predictive of early mortality.
    • Machine learning models can effectively predict mortality risk without relying on predefined severity thresholds.