High-performance Identification and Real-Time Monitoring of Calcium Metabolite in Urine with Calconcarboxylic Acid-Based Intelligent Detection System

The level of calcium metabolism in urine is closely related to homeostasis, reflecting the pathological changes of kidneys, bones, and thyroid. It is of great significance to build a convenient, visual, real-time, low-cost, and intelligent detection method suitable for routine health monitoring under home environment. Calconcarboxylic acid is selected as calcium-specific colorimetric probe and its activity, detection stability, and preparation reproducibility are ensured by using stable polymer carriers, and electrospinning technology. To overcome the nonlinear color transformations from the nonuniformity of euclidean distance (ED) distribution and the random error from every sampling point, deep convolutional neural network (DCNN) and smartphone app with highthroughput screening and excellent recognition capabilities, are used to analyze color signals on the chemosensors. When compared with ED method, the proposed DCNN-based method allows for sensitive detection of Ca2+ with 98 % accuracy by eliminating the loss of color extraction of sample and avoids limited sampling in ED method with 85 % accuracy. And the DCNN model is capable of extracting signal efficiently and detecting Ca2+ down to 0.1 mmol/L, well below the WHO regulatory limits, and wide linear range (0.1-3.0 mmol/L). The intelligent noninvasive detection system for high-performance identification and real-time monitoring of calcium metabolite in real urine is proposed for the pre-diagnosis and post-treatment monitoring of related diseases with abnormal calcium metabolism at home. Compared with the frequently-used flame atomic absorption spectroscopy, our proposed method is better for avoiding sample digestion, convenience, easy operation, and suitable detection performance under household conditions.