PhD Scientific Days 2023

Clinical Medicine - Posters A

Predicting higher energy need for effective defibrillation using machine learning in an animal model

Előadó neve

Dr. Pál-Jakab, Ádám

Neptun code

XH83LV

Előadó munkahelye

Heart and Vascular Centre – Semmelweis University

Előadó telefonszáma

+36309198410

Előadó e-mail címe

adam.paljakab@gmail.com

Az előadás címe

Predicting higher energy need for effective defibrillation using machine learning in an animal model

Szerző(k) neve és munkahelye

Adam Pal-Jakab MD1, Boldizsar Kiss MD1, Betta Nagy1
1. Heart and Vascular Centre – Semmelweis University

Bemutatás módja

Poszter

Szekció

Clinical Medicine - Posters A

Language of the presentation

Hungarian

Preferred session

Clinical Medicine

Összefoglaló szövege

Keywords: defibrillation threshold, efficacy prediction, arterial blood gas

Background Time is a critical factor in the medical management of resuscitation cases, requiring rapid, organized, and well-coordinated teamwork. While early defibrillation has been shown to improve treatment outcomes for cardiac arrest patients, the precise defibrillation strategy and the amount of shock energy required are still subjects of debate. This research aimed to analyze the database of an animal model of defibrillation (DF), assessing whether modern data-analysis techniques could facilitate the defibrillation energy selection process.
Materials and methods In the experimental setting, ventricular fibrillation was induced by 50 Hz direct current (DC), and then the defibrillation threshold (DFT) was determined using a step-down protocol. Arterial blood gas parameters (ABG) were measured before every defibrillation threshold measurement, levels of PaCO2, PaO2, pH, Hct, Na+ and K+ were recorded. The relationships between ABG parameters and the DFT were analyzed for 15 subjects using classical data analysis techniques and machine learning (ML) algorithms. Multiple ML models were trained and tested to predict the need for higher defibrillation energy need based on the ABG parameters.
Results Statistically significant differences were found in Hct and Na+ levels between two DFT categories, 130 Joules (J) and 40 J (p < 0.01). The DFT negatively correlated with PaO2 and positively correlated with Hct and Na+. However, other ABG parameters did not show significant correlations. Using ML, we predicted cases requiring higher defibrillation energy. Our best-performing model, the ExtraTreesClassifier, achieved 83% overall accuracy, with precision rates of 100% and 67% for higher and lower DFT categories respectively. We validated the model using bootstrap resampling and 10-fold cross-validation, both methods confirming consistent performance. We identified Hct, PaCO2 and PaO2 as significant contributors to model prediction based on the feature importance value.

Conclusion Our research highlights the potential of ML in guiding energy settings for effective defibrillation.

University and Doctoral School

Semmelweis University, Doctoral School of Theoretical and Translational Medicine

Supervisor

Prof. Dr. Endre Zima

Publication of my abstract

I do not give consent to the publication of my abstract on the website of the congress.

Kind

Szabad

Status

elfogadva

Accepted presentation method

poszter

Előadás fájl jóváhagyás

nem rendelkezett róla

Előadó

6901

Start

11:30

End

11:35

Authors (legacy)

Adam Pal-Jakab MD1, Boldizsar Kiss MD1, Betta Nagy1
1. Heart and Vascular Centre – Semmelweis University