Zeng Xinmu, Director of the Institute of Data Science and Engineering at National Chiao Tung University, said: "Big data, algorithms, and computing power drive the development of artificial intelligence. Initially, the development of AI operated in a way that knowledge was defined by experts, and machines simulated the thinking of experts to solve problems, and then machine learning methods appeared, and a new generation of algorithms such as Deep Learning were applied, along with high-speed computing. The subsequent emergence of machine learning methods, the application of a new generation of algorithms such as Deep Learning, with high-speed computing GPUs, to achieve AI systems such as Alpha go. The rapid development of AI medical care in recent years is the key to the integration of medical data is Big data, which will generate a very large force.
AI healthcare is considered one of the most promising fields by the venture capital community, and in the past two years there have been nearly 100 startups in North America, covering a wide range of fields, such as medical imaging, health risk prediction systems, etc. Google's AlphaGo began to cooperate with many hospitals after winning the Go tournament, for example, by applying it to diagnose diseases in the funduscopic retina, and the ITRI has a huge number of applications in the field of medical technology. Our ITRI Giant Information Technology Center and National Chiao Tung University have also cooperated to develop an automatic retinal lesion identification system, which collects tens of thousands of images and achieves a high recognition accuracy by means of a Deep Learning Model; and the integration of Medical Images and EMR data can further enhance the accuracy. Such a system can be used in frontline family medicine clinics to achieve the goal of early diagnosis and early treatment, and it can also be linked with laboratory physiological test data, and utilize the microvascular analysis of retinal fundus images to estimate the presence of cardiovascular diseases. Z In addition to technology development, Israel's Zebra has created a new business model in the AI medical field, allowing users to upload medical images to obtain a second doctor's advice service. Zebra not only collects a large amount of anonymized image data from around the world through this service model, but also allows users to pay for uploading their own data in order to obtain analysis services.
AI Healthcare Discovery AI healthcare finds more early lesions and improves the chances of early diagnosis ?
Based on the significant increase in AI image recognition capabilities, AI has recently begun to be applied to remote medical and care services, such as the U.S. Aicure platform using facial recognition and motion sensing technology, through the smartphone, the patient's automated observation of their medication and response behavior, to assist the physician in deciding on a prescription, to achieve remote treatment. The Aicure Platform Prof. Zeng Xinmu's team's iDeepCare project combines deep machine learning and huge amount of data analysis technology to develop an intelligent deep health care system. In cooperation with domestic medical institutions and experts, it collects all kinds of raw data on biomedicine and builds models, which can be practically applied to precision medicine, preventive health care, personalized medicine, risk prediction, etc. For example, the iDeepCare project cooperates with the Sanchong General Hospital to develop a colon endoscopic image development system using colorectal endoscopic images. For example, iDeepCare plans to cooperate with the Tri-Commission to develop an AI identification model of colorectal polyps using endoscopic images, which can assist physicians in identifying the nature of polyps, with an accuracy rate of 96%. In addition, iDeepCare is also cooperating with BeiJing in the development of an early warning system for arrhythmia, to predict the electrocardiogram performance of Bruegelder's Syndrome, which is associated with a high risk of sudden death. The AI machine can achieve 75% accuracy. Through the integration of technology to analyze large databases of hospital and health insurance information, it is possible to find the early foci of various diseases, and can dig out a lot of useful or unknown markers, which can help physicians make early diagnoses. This is the first time that we've seen this in action. Crossing the research threshold, AI will ***exist*** with humans ?
The biggest challenge for AI medical research is the quality of the data, how to avoid "Garbage in, Garbage out", and other thresholds such as cross-discipline cooperation, how to design business models, how to catch up with market demand, etc. Prof. Zeng Xinmu said that AI medical research can be based on existing data, and can be done through cross-discipline brainstorming. Prof. Zeng Xinmu said that AI medical research can think from the existing data base, but also through cross-field brainstorming to find innovative ideas, and the most notable is the application of instantaneous dynamic analysis, and the future cooperation between humans and AI*** will be able to significantly improve the medical accuracy rate, AI will not replace humans, but become an important helper for humans. The AI will be used for the first time in the future.