1. Diagnosing diseases: The biggest challenge facing medicine is the correct diagnosis and identification of diseases, which is also the top priority of the development of machine learning. A report of 20 15 shows that more than 800 kinds of cancer treatment schemes are in clinical trials. Using machine learning can make cancer identification more accurate.
2. Individualized medication: Using machine learning and predictive analysis to customize specific therapeutic potential for individuals is currently under study. If successful, this strategy can optimize the diagnosis and treatment plan.
At present, the focus of research is supervised learning. Doctors can use genetic information and symptoms to narrow the scope of diagnosis or make informed guesses about patients' risks. This can promote better preventive measures.
3. Drug development: Machine learning plays many roles in early drug discovery (such as new drug development) and research and development technology (next generation sequencing). The first item in this field is precision medicine, which makes the identification and possible treatment of complex diseases more effective. The clinical machine learning group of MIT is one of the main participants in promoting precision medicine by using machine learning, focusing on algorithm development.
4. Clinical trial: Clinical trial research is a long and arduous process. Machine learning can help shorten this process in various ways. One strategy is to determine the clinical trial candidates of the target population more quickly through advanced predictive analysis of a large number of data.
Analysts at McKinsey describe other machine learning applications that can improve the efficiency of clinical trials by simplifying tasks, such as calculating the ideal sample size, promoting patient recruitment, and using medical records to minimize data errors.
5. Radiotherapy and Radiology: Dr Ziad Obermeyer, an assistant professor at Harvard Medical School, said in an interview on 20 16: "After 20 years, radiologists will not exist in their present form. They look more like electronic robots: algorithms that supervise the reading of thousands of research reports every minute.
At present, deep mind Health of University College London Hospital is developing a machine learning algorithm to improve the accuracy of radiotherapy planning by distinguishing healthy tissues from cancer tissues.
6. Electronic health record: Support vector machine (technology for classifying patients' e-mail queries) and optical character recognition (technology for digitizing handwritten notes) are the basic components of machine learning system for document classification.
Application cases of these technologies include MATLAB of MathWorks (a machine learning tool with handwriting recognition application) and cloud vision API of Google.
A key point of clinical machine learning group of MIT is to develop intelligent electronic health record technology based on machine learning, and its concept is to develop "a powerful machine learning algorithm that is safe, interpretable, can learn and understand natural language from a small amount of labeled training data, and can be well popularized in medical environment and institutions".