Because under the current diagnostic methods, patients with knee osteoarthritis who experience delayed diagnosis are often unusual, they can only choose expensive and high-risk knee arthroplasty as a late treatment method in the later stage of the patient. Therefore, we use artificial intelligence to develop a community knee health warning system. This system collects low-cost data in community medical care, covering age, gender, height and weight, living habits, simple physical assessment results and related disease records, etc., and encourages community medical institutions to conduct early detection for the public. When a high-risk case is found, the system will remind the doctor to refer the patient to the hospital for early treatment in order to improve the treatment result, thereby protecting public health.