# Validation of intensive care unit predictive scoring systems in West Africans

## TL;DR

This study validates ICU scoring systems in West African populations to improve mortality and length-of-stay predictions.

## Contribution

The study is the first to validate ICU predictive scoring systems in a West African population.

## Key findings

- Existing ICU scoring systems will be tested for accuracy in predicting mortality and length of stay in Ghanaians.
- Supervised machine learning will be used to identify predictors of ICU mortality.

## Abstract

Intensive care unit(ICU) predictive scoring systems (PSS) are valuable in predicting outcomes in the ICU. However, they have not been validated in many populations, including West Africans, limiting their utility in these populations. The study population comprises Ghanaians managed in the ICUs of two major tertiary/quaternary hospitals in Ghana from 2017 to 2026. The Acute Physiologic and Chronic Health Evaluation-IV(APACHE-IV), Simplified Acute Physiologic Score-III (SAPS-III), Mortality Prediction Model-III(MPM0-III), Sequential (Sepsis-related) Organ Failure Assessment(SOFA)/quick SOFA(qSOFA), and National Early Warning Score-2(NEWS-2) scores will be calculated. For APACHE-IV, SAPS-III, and MPM0-III, the primary outcome measures are mortality and/or length of stay(LOS). The APACHE IV, SAPS-III, and MPM0-III scores and their corresponding predicted mortality ratios will be calculated. The overall observed and predicted ICU mortality rates will be assessed. Validation will be tested by assessing calibration and discrimination. For SOFA/qSOFA and NEWS-2, the associations of the mean score, highest score, and change in score with mortality will be assessed. Conventional statistical tools and supervised machine learning algorithms will be used to identify predictors of ICU mortality.•This study aims to validate the performance of existing ICU PSS in predicting ICU mortality and LOS among West Africans, and to identify their predictors.

This study aims to validate the performance of existing ICU PSS in predicting ICU mortality and LOS among West Africans, and to identify their predictors.

Image, graphical abstract