Intelligent Diagnostic System Development based on Artificial Intelligence Technology

Abstract

Intelligent diagnosis is an important scenario in smart healthcare, with conversational diagnostic scenarios being the most common. The process of collecting symptom information through conversations with users and inferring diseases based on symptoms. Through a dialogue based diagnostic system, it can meet some of the medical consultation needs of residents, thereby freeing doctors from some basic consultations and greatly alleviating the shortage of medical resources. In the actual diagnosis process, the symptoms reported by patients are often insufficient to support accurate diagnosis. It is necessary to ask the user if they have any other symptoms through dialogue to form a diagnostic conclusion. Existing research mainly adopts reinforcement learning methods, which gradually learn the dialogue process between traditional Chinese medicine students and patients in real medical scenarios, and obtain strategies for symptom inquiry and disease diagnosis. Despite the advantages of reinforcement learning in dealing with temporal decision problems, the diagnostic accuracy is still low and data dependency is strong. In this article, a medical dialogue robot architecture based on medical dialogue diagnosis technology, medical knowledge graph technology, and "inference machine" technology is proposed to build an intelligent diagnosis architecture. Secondly, in terms of algorithm, this article proposes a disease diagnosis algorithm based on Naive Bayes Classification and a symptom screening algorithm based on symptom set differences for symptom query process, This algorithm increases the interpretability of diagnostic results by simulating the questioning and diagnostic process of doctors, and combines it with the medical dialogue robot architecture to achieve intelligent diagnosis throughout the entire process.