Disease Outbreaks: Tuning Predictive Machine Learning

Abdullahi, T and Nitschke, G (2021) Disease Outbreaks: Tuning Predictive Machine Learning, Proceedings of Genetic and Evolutionary Computation Conference (GECCO 2021).

[thumbnail of 2021-Tuning Predictive Machine Learning.pdf] Text
2021-Tuning Predictive Machine Learning.pdf

Download (535kB)

Abstract

Climate change is expected to exacerbate diarrhoea outbreaks in developing nations, a leading cause of morbidity and mortality in such regions. The development of predictive models with the ability to capture complex relationships between climate factors and diarrhoea may be effective for diarrhoea outbreak control. Various supervised Machine Learning (ML) algorithms and Deep Learning (DL) methods have been used in developing predictive models for various disease. Despite their advances in a range of healthcare applications, overall method task performance still largely depends on available training data and parameter settings which is a significant challenge for most predictive machine learning methods. This study investigates the impact of Relevance Estimation and Value Calibration (REVAC), an evolutionary parameter optimization method applied to predictive task performance of various ML and DL methods applied to ranges of real-world and synthetic data-sets (diarrhoea and climate based) for daily diarrhoea outbreak prediction in a regional case-study (South African provinces). Preliminary results indicate that REVAC is better suited for the DL models regardless of the data-set used for making predictions.

Item Type: Conference paper
Subjects: Computing methodologies > Machine learning
Date Deposited: 03 Dec 2021 11:16
Last Modified: 03 Dec 2021 11:16
URI: https://pubs.cs.uct.ac.za/id/eprint/1488

Actions (login required)

View Item View Item