A remote computing based point-of-care colorimetric detection system with a smartphone under complex ambient light conditions.

Smartphone-based colorimetric detection has been one of the most commonly used techniques for point-of-care detection in recent years. However, there are two defects in the current detection system. One is the need of a light-tight box to isolate the impact of ambient light, and the other is the increased calculation with the number of probes. In this paper, a colorimetric detection system was coupled with a new color calibration method for detection under complex ambient light conditions. A 3 × 4 colorimetric probe array was used to display the color changes of different analytes. With the color calibration function and the Support Vector Machine discrimination function based on the RGB data captured at 5000 K preloaded in a remote server, the analysis results could be fed back immediately after sending the RGB data captured under complex ambient light conditions by the smartphone to the remote server. The discrimination results showed that this colorimetric detection system has a relatively high accuracy under complex ambient light conditions. An optimal probe selection algorithm (OPSA) based on the improvement of the traditional stepwise discriminant analysis was also proposed to dramatically reduce the number of probes for the identification of various analytes. The analysis results showed that this algorithm significantly simplifies the probe array and the simplified probe array kept exhibiting a good classification performance. Our research eliminates the dependence of smartphone-based colorimetric detection on light-tight boxes and cuts off the redundant probes, thereby greatly improving the portability of smartphone-based colorimetric detection.