Jaundice is describes as yellow discoloration of the skin and other tissues of a newborn infant. It happens when the bilirubin pigment rises up to 5 mg/dL or 85 μmol/L due the organs just started to developed. 75% of newborns in Malaysia had jaundice in the first week of life. Conventional assessment method is not convenient since it unable to detect all type of human skin colours, caused traumatization and costly. This research work proposed jaundice detection system based on the colour card technique that represent all types of human skin colours. This research does not involve any subject due to parent's cooperation and ethics limitation. The input data is represented by the colour card shades. It represents random bilirubin colour in patient's body, all types of human skin colour, and standard reference bilirubin concentration colour. It acquires input data by capturing different type of colour card shades using OPT101 photodiode sensor and USB4000-XR1-ES spectrometer. The input data collected were verified through Anderson-Darling normality test and analysed using Pearson's correlation coefficient to measure the strength of the association between the two variables for each input sample tested. Based on the correlation results, it shows high correlation value (r=0.997). Percentage error analysis is used to validate the experimental device and shows value of 9.90% which prove this technique is reliable. The significant contributions of this work is the improvement of the accuracy in detecting jaundice level for all type of human skin colours through this system.
