AI-Driven Decision Making for Auxiliary Diagnosis of Epidemic Diseases

Delay in diagnosis often leads to difficulties in the treatment of epidemics and disease spread. Therefore, early diagnosis plays an important role in the control of epidemic diseases. The rise of artificial intelligence(AI) technology provides more intelligent and effective methods for realizing auxiliary epidemic diagnosis. This paper first designs an auxiliary diagnosis architecture for epidemic, which supports data collection and processing for long-term monitoring of the target state. Then, using the iterative characteristics of time sequential decision, an auxiliary diagnosis decision-making model based on the partially observable Markov decision process is built to achieve early diagnosis of epidemics. Combined with state abstraction, a deep Q-learning auxiliary diagnosis(DQAD) algorithm is proposed to improve the timeliness and accuracy of epidemic diagnosis. Extensive simulations have been carried out to evaluate DQAD in terms of several performance criteria including average time per iteration and diagnosis accuracy. The result analysis verifies that the designed method is more accurate and reduces the diagnosis time than existing methods.