The number of people diagnosed with advanced stages of kidney disease has been rising every year. Although early diagnosis and treatment can slow, if not stop, the progression of the disease, many lower income individuals are unable to afford the high cost of frequent testing necessary to keep the disease progression at bay. To address this issue, we designed a kidney health monitoring system that allows for affordable and quick testing through the use of inexpensive test strips and a mobile application. Moreover, the application serves as a research framework for testing and improving detection models for the disease. In this paper, we describe the application we developed and several preliminary machine learning models we trained to classify the severity of the kidney disease as normal, intermediate risk, or kidney failure. We thoroughly evaluated the effectiveness of our models and found that our histogram of colors-based boosted tree method outperformed alternatives and exhibited good overall prediction performance (F1-score > 90%).