This AI application allows parents to detect signs of jaundice in babies from home.

In recent years, artificial intelligence has made a remarkable impact in the field of medical sciences, achieving significant advancements in areas such as cancer detection and drug discovery. Recently, an AI-driven application has been developed to ease the concerns of parents worried about neonatal jaundice, a common condition in newborns. This project is the result of collaboration between Singapore General Hospital, SingHealth Polyclinics, and the national health technology agency Synapxe.

The new application allows parents to assess jaundice in their newborn babies. It utilizes a machine learning model specifically adapted to Asian skin tones, showing encouraging results. According to the developers, the model achieved a sensitivity of 100%, performing favorably compared to existing testing methods for neonatal jaundice.

The application, known as BiliSG, was created after conducting two clinical studies over two years, during which multiple validations were carried out on infants under 21 days old. BiliSG analyzes the skin color in various areas of the body, such as the chest, forehead, and abdomen. Its name derives from bilirubin, a yellow compound that is produced by the breakdown of red blood cells and is also part of the digestive fluid produced by the liver. In cases of jaundice, bilirubin causes a yellowing of the skin due to its accumulation in skin tissues.

This tool allows parents to detect signs of neonatal jaundice from the comfort of their homes, thus avoiding frequent hospital visits. It is estimated that neonatal jaundice affects 60% of babies, with even higher rates in premature newborns. Generally, the body is able to regulate elevated levels of bilirubin in the bloodstream, but if it fails to do so, complications can result in permanent brain damage, developmental delays, and motor disabilities.

The medical research experts involved in the project state that this application will transform the way in which the brain health of babies is protected, enabling parents to monitor jaundice more accessibly, with the support and supervision of their doctors through telemedicine. To use BiliSG, parents only need to take a photograph of the skin area delineated by a color calibration label, designed to reduce glare, correct light intensity, and adjust to temperature variations.

Currently, the team is seeking to optimize the application by expanding validation. Initial tests were conducted using an iPhone 12, but they aim to extend its use to more smartphone models, including Android devices, which will involve significant effort due to variables such as camera resolution and lighting conditions. The goal is to make the BiliSG application more effective and reliable in real-world situations, as well as to simplify its use for parents and caregivers who are not familiar with medical tools.