What can AI systems do in medicine?
Tasks that are not directly related to the treatment of patients
Alina Testova, Habr
For links, see the original article
Modern artificial intelligence systems are already helping doctors treat patients. For example, the HeartFlow company, using CT scans, computer modeling of blood flows and deep learning algorithms, is able to build a 3D heart map. This gives doctors the opportunity to diagnose heart diseases more accurately and faster, reducing the number of necessary invasive procedures by 80%.
However, AI is also used in areas that are not directly related to the treatment of the patient, but still affect the quality of medical care. Today we want to talk about these, to some extent auxiliary, but still important tasks.
Routing in hospitals
Artificial intelligence systems and machine learning can help not only in making a diagnosis. For example, at the end of May, the University College London Bloomsbury Clinic (UCLH) announced that it would use AI systems to identify patients who really need emergency medical care.
When a patient complaining of pain arrives at the emergency room, the medical staff will perform standard procedures – take blood for analysis, collect anamnesis, if necessary, do an X-ray. As noted in the polyclinic, in 80% of cases, patients have nothing serious – they are prescribed medications and allowed to go home.
The artificial intelligence system will allow you to quickly identify the very 20% who really need urgent care. The CEO of UCLH told the Guardian in an interview that the software will set a priority for the patient, assessing the danger of the symptoms voiced by him. For example, abdominal pain can mean appendicitis or kidney disease, so such a person will "move" towards the head of the queue.
Machine learning algorithms can also help with routing patients and doctors. For example, a researcher and consultant neurologist at the National Hospital of Neurology and Neurosurgery of Great Britain, Parashkev Nachev, has developed a machine learning algorithm that analyzes information about clinic appointments and estimates the likelihood that a patient will miss an MRI scan session for one reason or another. Its system takes into account such parameters as a person's age, address and distance to the clinic, weather conditions. So far, the scientist has managed to achieve an accuracy of 85%. This helps to quickly redistribute the recording time.
And in the same UCLH, the artificial intelligence system, which is being developed by scientists from the Alan Turing Institute, will track how doctors and patients "move" around the hospital – what tasks they perform, what procedures they go to. This will help identify potential bottlenecks in the organization of polyclinic work – situations or places where queues or equipment shortages are potentially possible.
Search for new knowledge
The treatment practices followed by doctors tend to become outdated. New methodologies, new research and drugs are emerging. Back in 2004, researchers studied the contents of 341 medical journals and found that the total number of monthly publications exceeded 7 thousand.
Ideally, a doctor should constantly maintain the level of subject knowledge, be aware of modern treatment practices – however, it is almost impossible to study the entire body of publications that are regularly published in thematic journals – even if we are talking about a narrow specialist.
Artificial intelligence technologies in combination with search engines can help in this situation. A similar solution was developed by scientists from the American RAND Research Center, which deals with methods of analyzing strategic problems. They taught the system to search in huge amounts of information for keywords and terms related to the subject of the query.
During the tests, this topic was data on gout, low bone density and osteoarthritis of the knee joint. The algorithm managed to reduce the number of relevant articles of interest to doctors by 67-83%. According to the developers, the system missed only two articles that would have been selected by people, but none of them contained critical information. The accuracy of the machine learning algorithm was 96%.
Drug development
The experience of pharmaceutical companies shows that approximately 12 years pass from the start of preclinical trials to the approval of the drug and the treatment of patients. At the same time, only 0.1% of "candidate drugs" get to clinical tests. Approval is received by 20% of them.
Artificial intelligence systems can help resolve this situation and accelerate the release of new drugs. Machine learning and AI systems are used in the early stages of drug development.
An example is the solution of AtomWise company from San Francisco. Their system is called AtomNet. It uses deep learning techniques to predict how molecules will behave and with what probability they will form the necessary bonds.
During the training, AtomNet developers "fed" the artificial intelligence system data on the results of several million already known interactions of molecules. Based on these interactions, the system has learned to predict interactions that have not yet occurred. The software has already helped to create drugs for the treatment of Ebola.
Artificial intelligence systems and machine learning help doctors and scientists work more efficiently. Doctors are freed from routine tasks, it becomes easier for scientists to conduct research, and patients receive treatment faster.
Today, developments at the intersection of AI and medicine are becoming increasingly popular. For example, Google began selecting companies engaged in the creation of "medical" artificial intelligence systems to participate in the Launchpad Studio startup accelerator program. At the end of last year, four companies joined the project at once.
We at DOC+ are also engaged in developments in this area: we are developing our own NLP system that can process texts on medical topics. It is used in our chatbot - it helps to collect anamnesis, is able to isolate the symptoms of diseases from the patient's complaints and transmits them to the doctor in a structured form.
Portal "Eternal youth" http://vechnayamolodost.ru