The present work proposes a low cost and portable Non-Invasive Blood Glucose Measurement (NIGM) system based on Near-infrared (NIR) light. In vitro system using single LED (940 nm) with an array of photodetectors is fabricated. Regression analysis is carried out to study the relationship between detector output voltage and actual glucose concentration. Low RMSEC (reflectance: 12.87 mg/dl, transmittance: 15 mg/dl) of in vitro measurement motivated to design a sensor patch for non-invasive in vivo glucose measurement. The accuracy of our indigenous device was tested by comparing non-invasively estimated blood glucose with invasively measured blood glucose. To estimate the glucose concentration from the detector voltage signal an ADAptive Linear NEuron (ADALINE) based Neural Network structure is used. The calibration model is prepared using data of 10 non-diabetic subjects. The observed RMSEP was 14.92 mg/dl with correlation coefficient (0.87) in the case of In vivo experiment. The prediction of glucose concentration is in the clinically acceptable region of the standard Clark Error Grid (CEG). The proposed design of NIR light based glucose measurement can be used to develop an NIGM system. © Springer Nature Singapore Pte Ltd. 2018.