Small Robots And Big Money: What’s The Basis of $98 Bln Future Healthcare
The healthcare market is preparing for a new technological revolution due to the introduction of neural networks and robotic systems. According to a study by Frost & Sullivan Agency, the share of artificial intelligence in healthcare will grow by 40% annually, and use of “smart systems” revenues will raise to $6.7 billion by 2021, meanwhile the accuracy of AI diagnostics will increase by 30-40%.
As for the use of robotics as a type of minimally invasive surgery, according to the Allied Market Research the surgical robotics market in 2017 was estimated at $56 billion, and with prospects of growth annually by 8.5%, it is expected to reach $98 billion by 2024. Basically it’s about the implementation of miniature surgical instruments to perform precise manipulations during surgery procedures.
In an interview with Bitnewstoday.com Maria Victorova, PhD candidate, developer of robotic systems for medical diagnostics from Department of Biomedical Engineering of Hong Kong Polytechnic University, told about the latest R&D trends of robotics in healthcare.
BNT: When people talk about robots in medicine, they often think about Da Vinci robotic surgeons.
MV: Indeed, it’s a very sensational story about Da Vinci robotic systems, which are still leading in the market, holding the first position by the number of performed operations, so clinics are increasingly welcoming their installation. The reliability of the operations carried out by these robots guarantees peace of mind for doctors, and this case stimulates the launch of a new generation of Surgical robotics.
At the time of its creation in 2000, Da Vinci robotic system was revolutionary. Its cost, which ranges from $1 million to $2 million, not including maintenance costs of $150 000, pays off for 2 years on average. At the same time, from the time of their launch, the only 5 to 10% of all surgery procedures are performed with their help among the total amount of operations in the world. The main reasons of that are their cost, as well as the volume of the device, as the 750-kg robot is not designed for frequent movement by the operating rooms. To solve this problem, researchers from Cambridge Medtech CMR Surgical launched a surgical robot Versius, which investors, including LGT and ABB Robotics, spent on it $95 million in June 2018. It is not clear yet whether the new digital surgeon will be able to revolutionize medical robotics, but its inventors hope that it can become more economical and convenient solution
BNT: What are the main trends in robotic healthcare systems in general? What is the focus of research and development?
MV: The robots that are now emerging in the industry are mostly directed towards collaboration with a doctor that operates them. According to their standards and internal characteristics, they are completely safe and fit to cooperation with environment and humans. This is the reason why many researchers have started to work on medical applications of existing inventions to meet robotics standards. And there is a significant surge in research of the new technology in its application in the diagnostics: surgical robots are too expensive, while diagnostic will be more available for mass use.
BNT: What are the main benefits of using robots in medical diagnostics?
MV: They can eliminate human error and incompetence, reduce errors and make the process more or fully automated. Accordingly, there will be less demand for doctors specialized in the narrow field of medicine. This doesn’t mean that it will necessary replace the doctors or even reduce the number of jobs. It will simply be enough to take a person with less competence who can assist the robotic system and manage it.
Testing of robotics in medical diagnostics is quite active. The neural network GoogLeNet, showed the accuracy of breast cancer recognition by 89% in a couple of minutes, while an experienced pathologist, working without time limit, determines it with 73% accuracy. But despite of that, scientists are not ready yet to completely exclude the doctor from this scheme, since AI is much worse at detecting benign tumors.
Scientists are not ready to give the final medical report to the AI hands yes, for many reasons. In comments to Bitnewstoday.com Andrei KVASOV, PhD computer science student of Hong Kong University of Science and Technology, called it an “AI safety problem”: "When it comes to human lives, in addition to ethical issues - like who must be entrusted to take solutions - a human or a machine, and who will be responsible for it… the big question of the accuracy of algorithms is very acute.
BNT: How to find out that the diagnostics by a robotic AI system was accurate and why they are not used everywhere?
MV: I can't tell any figures, but there are definitely difficulties in several aspects. Firstly, where to find enough data for AI training to achieve great accuracy? Each medical institution works with its own databases, and one common storage system could solve this problem. However today, the security concerns people more than science.
Andrey KVASOV confirmed that today the issue of the accuracy and autonomy of AI in the diagnostics is still under the great question. "Neural network algorithms were developed about 50 years ago, but there is still no direct explanation on what principle they form patterns and correlations. So, we have no guarantees of their accuracy for now.” According to the expert, it is also difficult to adapt the algorithm to another disease detection. And even high accuracy of identification of one disease doesn’t allow to define another one by means of the same neural network. Since training is entirely dependent on the loaded data, new information will be required for the AI to be able to analyze another disease. That means that neural network will actually need to be trained from scratch.
MV: The second problem preventing the spread of AI diagnostics is rather an ethical one: if a doctor falsely diagnoses a disease based on a neural network analysis, looking at the patient's disease as if “from the tip” of the AI, who will be responsible in such case - a doctor or a machine? Not mentioning the fear of losing jobs and replacing doctors by the neural network, which is impossible at this stage anyway.
Back in 2016, a team of Harvard scientists has achieved 85% accuracy in the breast cancer detecting, combining the forces of pathologist and neural network, which made it possible to distinguish cancer cells from the healthy ones. And in September 2018, the Mayo clinic taught artificial intelligence to determine the level of potassium in a body with an electrocardiogram, without a blood test. Food & Drug Administration (FDA) has already announced that this startup is already on its way to clearance status and implementation in the healthcare system, but there is no final decision yet
MV: In my opinion, the solution for all these problems lies in the parallel and independent assessment of the disease by several experts, one of which the AI is actually becoming. Thus, it’s about a collaboration, as a result of which more accurate diagnosis is going to be possible.
BNT: Could you tell about some successful use-cases of robotic diagnosticians?
MV: One of the main reasons for the robotics popularity in diagnostics is their use in ultrasound research. They are absolutely safe for the patient and allow to avoid using x-rays methods, that are not so harmless - for example, Xray, CT. And among the other things, automation of this process also takes care of doctors, reducing the number of their actions and preventing their musculoskeletal disorders. Another reason for the popularity of these robots in the diagnostics is the ability to carry out the survey distantly: the doctor can remotely check the status of patients in several hospitals. This is the so-called teleoperation, when the doctor makes an inspection, being at any distance from the patient and controls the robot with a joystick.
However, despite the huge number of developments in this range of medical services, there is still quite a small amount of commercially available versions of robotic ultrasound systems: ABUS by Siemens Healthcare - for breast research and Kuka by LBR Med.
BNT: What developments does the industry focus more on - surgical or diagnostic robotic systems?
MV: The industry's interest in robotic ultrasound diagnostics is huge, which is due to their main advantage over surgical robots: while the last ones specialize in a particular application and their variation totally depends on the type of operation, the diagnostic robots can be adapted to different tasks and inspections. So, changing them to a minimum degree - they will differ only in the settings of force, angles and trajectories. And since the diagnosis is needed by more people than surgery, it is very important now to make this method more accessible.
Full autonomy of robots in medicine is extremely difficult to achieve due to a number of reasons. However, researchers are actively testing the technology. Thus, in June this year scientists of the Oxford center for biomedical research with the support of the National Institute of health research conducted the first successful surgery using robots on the retina at John Radcliffe hospital
BNT: What kind of research does your institute carry out in this field?
MV: Hong Kong Polytechnic University has developed medical equipment that creates a 3D ultrasound image of spine – Scolioscan. This device also avoids x-rays, which is especially important in the treatment of children, as they have to do it often when they are in the process of spine treatment and need a constant monitoring. Now we are working on using a universal robot to automate scanning and navigation of equipment, completely eliminating the need for the presence of a doctor in this procedure.
BNT: When talking about the future healthcare, it’s often the idea of non-invasive gadgets injected into the human body without surgery that occurs in mind. How relevant such concept is and what promising developments are being carried out in this direction?
MV: Indeed, there are robots that are injected directly into human body in different ways. Researchers at Chinese University of Hong Kong (CUHK) are developing special endoscopic capsules with automatic diagnosis. These tiny devices use the ML-based camera records in the gastrointestinal tract to determine the problem area, and then analysing the intestinal tissue, they can identify anomalies.
There's also an interesting R&D at MIT - the origami robots of microscopic size, controlled by a gradient magnetic field. Having fulfilled its function in the human body, they biologically disintegrate, leaving just a small magnet, which is excreted from the body by natural way. It can be argued that this intervention is the minimum load on the human body.
In the next article, we will continue our interview with Maria Victorovaand talk about the use of AI in healthcare in such promising and innovative technologies as AR and VR.
According to the Allied Market Research report, the world leader in surgical robotics in 2017 was the US. According to forecasts, they will retain the dominance until 2024, due to the well-established health infrastructure and the increase in the introduction of robotic surgical systems in medical institutions. Spain is expected to be the fastest growing country in the European market, and India is expected to be a leader of Asia-Pacific region. In general, according to the study, the services of surgical robotic systems will become a profitable industry - it is expected to increase annually by 15.7%. At the same time, neurosurgery, for which the utmost accuracy during surgery performance is so much important, will be the largest and most expensive amounting to 1/6 of the entire market of robotic systems
Back to the list