AI is seen as a possible silver bullet for medicine and healthcare. But can AI truly transform these realms? The answer is no in the short term, but very likely in the long term.
Right now billions of dollars are being invested in AI research for medicine, medically oriented human biology, and health care. It is not surprising why upon the waves of myriads of sensational headlines from media outlets, we are becoming more and more intrigued about what results it can bring.
Machine Learning, Deep Learning, Neural Networks… the main directions of AI that are increasingly finding its practical application in creating data processing systems and automated diagnosis of diseases, in developing new drugs based on modeling, and also in developing systems for analyzing various types of biomarkers. But, is it really workable?
Some of these soar in the hype and deliver, and some crash in delivery proving a false hope. So where does Artificial Intelligence fall with respect to medicine?
Ex Medica: Ways AI is Really Affecting the Medical Field
For the majority of us, AI is a technological miracle-worker and naturally, it became the focus of a hype cycle. It pushed itself into the spotlight as the latest revolution in the medicine to solve different problems (or pretend to apply it?).
So, AI and medicine: greatest hype or hope? Let’s review the summary of all the latest movements that will help to answer this question.
*I made this list on the basis of facts and real achievements, no yellow journalism. All the links to the sources are attached, so you can read more info on each innovation.
Successful treatment of cancer primarily depends on their timely diagnosis, so research in this area is very relevant. Here are the most significant AI contributions to cancer diagnosis and treatment:
- Google-team researchers presented an Augmented Reality Microscope at the annual conference of the American Association for Cancer Research, AACR. As a result of testing, the prototype learned to identify lymph node metastases with a probability of 0.98, and prostate cancer with a probability of 0.96 — that is, with an accuracy of 98 and 96 percent, respectively.
- The Chinese technology corporation Baidu has developed an artificial intelligence system for diagnosing cancer, as well as assessing the possible development of the disease in a patient’s future.
- Higia Technologies developed EVA — a biosensor insert into a bra that thermally reads data that is subsequently transmitted via Bluetooth to the app, and AI analyzes the results to provide the user with an estimate.
- Scientists from Osaka University have presented a system that can detect various types of cancer cells by scanning the images obtained with the help of microphotographs with an accuracy that exceeds the capabilities of a person.
Analyzing these achievements, one can hope that stable, fully-funded research, the use of the latest medical equipment, and most importantly, cutting-edge technology of AI, will help bring victory over this terrible disease closer.
AI-developments in the field of Cardiovascular disease (CVD) are no less promising. Right now AI impresses with its results in predicting and treating cardiovascular pathologies and fully justifies the considerable financial resources invested in it. Here are the main movements worth noting:
- German engineers Nils Strodthoff & Claas Strodthof have developed a neural network that can detect signs of myocardial infarction. They provided an algorithm that operates directly on ECG data without any preprocessing and to investigate its decision criteria.
- A team at the University of Auckland’s Bioengineering Institute have created a virtual 3D heart that can provide a huge breakthrough on the treatment of the most common heart rhythm disturbance, atrial fibrillation (AF).
Of course, it is not the full list of achievements. In general, over the past years, engineers and scientists have achieved quite good results in the diagnosis and treatment of CVDs. As a matter of fact, timely detection of myocardial infarction and other serious heart diseases could save hundreds of thousands of lives. Consequently, AI research in this realm is very important.
As the science of the nervous system and the brain, neurobiology is quite a new direction in medicine, but it has already found wide application in many directions. The use of AI here seems justified and invaluable, and the painstaking work on the creation of new developments over the past year speaks of this. Just check out the most significant achievements:
- A group of scientists at Stanford University has developed a three-dimensional convolutional neural network (3D-CNN) for the diagnosis of Alzheimer’s disease, its prodromal stage, weak cognitive impairment (MCI) using MRI images. The study revealed a hippocampal region in the brain that is critical for the diagnosis of Alzheimer’s disease.
- Researchers at the University of Manchester created the SpiNNaker ‘Human brain’ supercomputer with 1 million processors (Spiking Neural Network Architecture), which allows simulating high-level real-time processing in a range of isolated brain networks. It also has simulated a region of the brain called the Basal Ganglia — an area affected in Parkinson’s disease, meaning it has massive potential for neurological breakthroughs in science such as pharmaceutical testing.
- MaxQ AI company has powered up the Accipio Ix system, which is capable of detecting intracranial and intracerebral bleeding using CT without contrast. This program can be easily integrated into the tomograph and the PACS * system using the DICOM industry standard.
To sum it up, AI achievements in Neurobiology have small yet practical value. Let’s hope that the creation of a supercomputer “SpiNNaker will delight us with magnificent achievements very soon.
Predictions and Diagnose diseases
Without any doubt, successful prediction of such serious diseases as cancer and cardiovascular diseases can significantly increase the percentage of human survival. In this area, one of the most interesting and important AI-powered developments is an algorithm presented by researchers from the University of Nottingham, which is able to analyze the patient’s condition and predict death.
In addition to assessing the risk of death, this technology helps to build different treatment options for patients, determine the period of stay in the hospital, the likelihood of repeated visits to doctors, process patient data, such as notes of doctors in medical records. Plus, the neural network analyzes the available information and gives its assessment, making conclusions faster and more accurately all existing technologies.
Another achievement worth attention is an algorithm that is learning to detect whether patients will wake from a coma. The AI algorithm, developed by the Chinese Academy of Sciences and PLA General Hospital in Beijing, analyzes fMRI scans of a patient’s brains to gauge how blood flows to different areas of the brain, as well as information given by doctors like the patient’s age, how long they’ve lost consciousness, and the cause of the coma, and then makes its diagnosis.
A particularly important direction in medicine is the search and creation of drugs. Machine learning algorithms come to the rescue in finding new compounds and analyzing chemical reactions. One of the most promising research is a system developed by researchers at the University of Glasgow, it monitors substances that arise during the reaction and disappeared using infrared spectroscopy methods based on nuclear magnetic resonance in real time.
What is more, a group of scientists from Stanford University has developed a system powered by artificial intelligence, which can analyze the interaction of two drugs. At the moment, the system is analyzing the interaction of no more than two drugs, but scientists expect in the future to expand them to more complex drug combinations.
An interesting project is a new molecule design strategy with target properties, called ReLeaSE (Reinforcement Learning for Structural Evolution), presented by American researchers. The algorithm is comparable in quality with other means of chemical modeling and allows, for example, to predict the coefficient of distribution of the molecule in the octanol/water system with an accuracy of 91%.
AI is getting closer to replacing animal testing and probably there is nobody who will not be excited about this news. A team of researchers from Johns Hopkins University has developed a computerized approach to detect toxicity of chemicals, exceeding the accuracy of animal experiments. In such a way, we can save more animals lives and made our world a better place!
Final Verdict: Is AI Promising More Than It Can Deliver?
As a crucial part of our lives today, whether you realize it or not, AI assists in almost every scenario. Medicine and healthcare are also significant parts of this, and due to the myriads of sensational headlines, we can probably call it an overhyped theme. However, the truth is something in the middle.
Undoubtedly, AI is making little yet confident steps to transform medicine. Right now it can more efficiently diagnose diseases, develop drugs, personalize treatments. What is more, it can easily save doctors from doing tasks like taking notes and reading scans, allowing them to spend more time connecting with their patients.
As I see this situation, it seems that science is moving in the right direction. This is just the beginning, and I want to believe that barriers to a long and healthy life will gradually be overcome by science.