Evaluating Biomedical Features for Reactive Hypoglycaemia Prediction with Machine Learning

DOI

10.2991/978-94-6463-112-8_11How to use a DOI?

Keywords

Reactive hypoglycaemia (RH); machine learning (ML); biomedical; cohort

Abstract

Reactive hypoglycaemia (RH) is a condition when post-prandial plasma glucose level drops, usually <70 mg/dl. Its exact cause is unknown and associated with risk of diabetes among non-diabetic individuals. Attempts are made to predict RH but met with difficulties due to varied data characteristics and inconsistent results among different studies. Hence, machine learning (ML) techniques serve as alternative to improve model accuracy on condition that more data are added. This study was aimed to evaluate ML algorithms in RH prediction based on limited data of biomedical features. Data of 1540 participants of biomedical cohort study of risk factors for non-communicable diseases in 2021 were selected. Binary RH was set to be target variable. Analysis was carried out using Orange software with five supervised ML algorithms, i.e., logistic regression, decision tree, random forest, support vector machine (SVM) and gradient boosting. The result showed that RH cases were found at 1.36%, one among these being diabetic and two had prediabetes. SVM gave the overall best performance with area under curve (AUC) of 0.733 compared to other algorithms. However, classifier evaluation metrics (F1, precision, recall) were much better if non-RH condition was selected as the target. RH prediction unexpectedly plunged their respective values to zero except for gradient boosting (F1 0.027, precision 0.060, recall 0.018), indicating large difference of samples between two categories of RH variable. As promising as it is, these results suggest that careful interpretation of ML-based modelling is still mandatory when discrepancy of sample size between classified groups is encountered.

Copyright

© 2023 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Frans Dany
AU  - Fifi Retiaty
PY  - 2023
DA  - 2023/03/01
TI  - Evaluating Biomedical Features for Reactive Hypoglycaemia Prediction with Machine Learning
BT  - Proceedings of the 1st International Conference for Health Research – BRIN (ICHR 2022)
PB  - Atlantis Press
SP  - 105
EP  - 113
SN  - 2468-5739
UR  - https://doi.org/10.2991/978-94-6463-112-8_11
DO  - 10.2991/978-94-6463-112-8_11
ID  - Dany2023
ER  -