Integrated Rapid Test with Artificial Intelligence for Total Bilirubin Quantification in Blood Samples
摘要
Bilirubin is an important marker of hepatic metabolism. Through it, liver function problems can be identified, characterized by an increase in bilirubin concentration, which can lead to a condition known as jaundice. For disease diagnosis and quantification of the analyte, the biochemical method is currently the reference method. With the advancement of science and its technologies, other methods for bilirubin quantification have been developed, including rapid point-of-care alternatives. In this work, the development and validation of a rapid test for total bilirubin quantification in human samples, including neonates or adults, are reported. For this purpose, a prototype was developed with the capacity to filter whole blood so that the analyte could be quantified in the liquid fraction (serum). The quantification of the analyte is possible through the reading of the prototype on a commercial device, which, through the standardization of the embedded AI, can determine the concentration in the sample after an extensive training process to acquire a better correlation between the AI diagnostic and the bilirubin concentration. Preliminary results showed that the prototype was able to differentiate bilirubin concentrations from 1 to 7 mg/dL, thereby distinguishing between normal and pathological concentrations. Some experiments showed a miss of accuracy of 0.3 mg/dL in quantification, indicating a point for improvement in the development process. Seeking information in the literature, similar rapid test devices exhibited behavior close to that found in the development of this device. As the next steps, studies will be conducted on the device used for quantification, as well as different methods of data extraction from the prototype to standardize the quantification model.