With the increasingly mature AI (Artificial Intelligence) technology changing into “Ai +” in all walks of life, “Ai + medical”, as a field that can produce epoch-making changes and is directly related to human well-being, has naturally become the focus of many giants. Although the current specific applications of “Ai + medical” mainly focus on the fields of machine learning assisted diagnosis and analysis, compared with the artificial intelligence industries such as driverless cars limited by technology and law, intelligent medical is obviously easier to land and has the conditions to become the next “Ai blue ocean”.
At present, artificial intelligence has been widely used in the field of medical and health. From the perspective of application scenarios, it is mainly divided into four aspects: virtual assistant, medical imaging, drug mining and nutrition. With the gradual improvement of speech recognition, image recognition and other technologies, the pan artificial intelligence medical industry based on these basic technologies has also become mature, which has promoted the rapid development of the whole intelligent medical industry chain and the birth of a large number of professional enterprises.
1、 Virtual Assistant: a right-hand assistant or a substitute for human doctors?
In the medical field, the virtual assistant can intelligently judge the etiology through the disease description according to the conversation with the user. Therefore, virtual assistants are mainly divided into two categories, one is a general-purpose virtual assistant including Siri, and the other is a special virtual assistant focusing on medical and health. General purpose virtual assistants are listed early, with high capital support and large data scale. The medical virtual assistant has strong professional attributes and high regulatory risk.
Virtual assistant is a more popular artificial intelligence medical and health subdivision field favored by capital. At present, the medical and health virtual assistant familiar to users is Babylon health.
At present, the regulatory authorities require that the virtual assistant can only provide some advice and suggestions in light diseases, but can not provide diagnosis. In severe diseases, it can only propose to go to the hospital immediately or dial the emergency telephone of the hospital. Doctors in the industry also have some doubts about the application, because patients do not fully understand the situation of the body, and some key information will be omitted during expression. At the same time, a large number of non professional words will be used during consultation. The virtual assistant may not be able to mine the really useful information and make a more accurate judgment.
These are the current problems of virtual assistant. Nevertheless, the cost of virtual assistant is lower and helps to control expenses. Human doctors can’t exhaust all diseases. In theory, artificial intelligence can, so it can become a right-hand assistant for human doctors. For the future, with the rapid development of machine learning and the intellectualization of medical detection methods, many people are full of hope that virtual assistant can replace human doctors.
2、 Medical imaging: assisting and replacing doctors to see films
The combination of medical imaging and artificial intelligence is a new branch and industrial hotspot in the field of digital medicine. Medical images contain a large amount of data, even experienced doctors sometimes seem at a loss. The interpretation of medical images requires the accumulation of professional experience for a long time, and the training cycle of doctors is relatively long. However, artificial intelligence can do faster than professional doctors in terms of image detection efficiency and accuracy, and reduce the misjudgment rate of human operation.
In recent years, the performance of image recognition technology has been rapidly improved with the help of “deep learning”. In the process of AI assisted diagnosis, AI will also make in-depth learning and find cases in the medical record database as the basis for judgment.
In medical imaging enterprises, the addition of artificial intelligence technology also has a great impact on the core competitiveness of entrepreneurial teams. According to the investigation of medical imaging start-ups by research institutions, with artificial intelligence technology, the whole team can significantly reduce labor costs. If there is no artificial intelligence technology, it is necessary to form a customer service team with high labor costs to communicate with doctors. The ratio of technicians to non technicians is 1.1:1, and the scale has reached 30-50.
3、 Drug Mining: greatly reduce the cost of drug R & D
Generally, the development of a new drug takes an average of 10 years and costs billions or even tens of billions of dollars, which is also one of the important reasons for the high cost of drugs. However, artificial intelligence provides people with a safety expert to detect drugs at a lower cost.
First, in the screening of new drugs, several candidates with high safety can be obtained. When many kinds or even thousands of compounds show some curative effect on a disease, but it is difficult to judge their safety, the search algorithm of artificial intelligence can be used to select the best candidate for new drugs.
Secondly, artificial intelligence can also be used to detect the safety of new drugs that have not yet entered the stage of animal experiment and human experiment. Artificial intelligence can screen and search the side effects of existing drugs, so as to select those drugs with the least probability of side effects and the least harm of side effects to enter animal experiments and human experiments, so as to save time and cost.
4、 Nutrition: machine learning gives you more accurate and personalized nutrition suggestions
By analyzing the results of standardized diet, medical experts found that even if they eat the same food, there are still great differences in the reactions of different people. This shows that the “recommended nutritional intake” derived from experience in the past is fundamentally flawed. Based on the fact that blood glucose management is the cornerstone of precision nutrition, the researchers developed a set of machine learning algorithms to analyze and learn the relationship between blood samples, intestinal flora characteristics and postprandial blood glucose level, and tried to predict blood glucose with standardized food.
The results show that the machine learning algorithm can give more accurate nutritional suggestions and successfully control the postprandial blood glucose level, and the results are better than the traditional expert suggestions. Reasonable dietary collocation and the demand for safer organic food have become the new growth point of the food industry, and have rapidly become the traditional field promoted by new technology.