A Machine Learning Based Gestational Diabetes (Gd) Prediction

Abstract

Machine Learning (ML) communication technologies have brought immense revolutions in
various domains, especially in health monitoring systems. Machine learning techniques
coupled with advanced artificial intelligence techniques detect patterns associated with
diseases and health conditions. This paper presents a non-invasive breath test to monitor the
condition of diabetic patients. It is identified as an easier technique and quick diagnoses of
diabetic ketoacidosis (DKA). DKA is a preventable acute complication of type 1 Gestational
Diabetes (GD) mellitus. Common diabetic test on patients are done on urinary test and blood
ketone test to monitor for Gestational Diabetes (GD) condition. However, those methods are
considers as invasive, inconvenient and expensive. Recently, breath acetone has been
considered as a new ketone biomarker because it is non-invasive, convenient, and accurate
reflection of the body's ketone level. This research presents a method of monitoring ketone
level by using breath measurement. Main objective of this research is to present an easy
handheld health care on monitoring diabetic level. Method consists of development of
hardware connection with Machine Learning (ML) system to facilitate the process of patients'
diagnosis and personal monitoring. In this system, Arduino board is used to read the sensor
with sense the breath. Breath value level is log to system using wireless communication. Data
collection is interfaced to web page. Ketone level is measured as the amount of breath
acetone is collected when patients exhale into a mouthpiece that consists of gas sensor. This
research is significant where patients can independently monitor their diabetic health and the
ML system can be alerted directly to medial officers in the hospitals.