The questions about AI implementation in different industries are often raised in regard to human labour force optimization and substitution. The medical and healthcare industries are also subject to rapid change in machine learning technologies and seem to be ones ripe of change. There are already certain applications in the hospital or care environment where the participation of an AI is a more effective option — from data administration to security and processing. There are unfounded concerns that, with further developments, machine learning and robotic systems will take away the jobs of doctors and other medical personnel. This is often exaggerated in media outlets, so let’s take a look at the issue from different angles to draw up a short-term conclusion based upon recent developments.
The problems that doctors encounter on a daily basis
As a former doctor, and someone who observes many practising doctors, I believe that our roles are going to become less focused upon analysis and that is due to the increasing role of machine learning. This is not a bad thing: every doctor will tell you that the strain upon healthcare systems makes it difficult to concentrate upon the important aspects — like getting to know the patient as a person rather than just a set of problems that need to be solved. The healthcare industry and the role of a doctor is as much about the data analysis as the caring role, where the most essential part is the interpretation of the information presented to map the needs of the patient out. A lot of the duties of a typical doctor can be automated and made more accurate. Think of the life of a busy doctor in your local hospital and you will realise they spend a lot of time digging into diagnostic information, examining blood test results, chemical results, looking at X-rays, MRI, trying to find causality in a person's medical history — making difficult, rapid decision under risk and uncertainty.
Looking into the future — let's say 5 years, this process can be automated by a hybrid of blockchain technology, machine learning and big data analysis (statistics) to move trusted data around the multidisciplinary team — regardless of location — and utilising AI to analyze that data. Healthcare will be borderless, data flows will be less constrained but, importantly, the democratic oversight will still be there to give the patient control of what their data is used for.
A company I’m involved in, is trying to open up participation so that the cost of healthcare is equalised by machine learning — by allowing patients to select a location for their healthcare needs that suit their budgets regardless of where they live. Another startup that I am working with is incredibly ambitious — they want to create an entire industry that unites preventative medicine, health and fitness, performing arts, mental health, the insurance industry and social media via machine learning. Their CEO has often said “to get people to be healthy we need to get them to move or exercise in a way that is helpful to them — not what their peers or social media account tells them to do. We need to base decisions on data science to help reduce the burden upon our public healthcare systems.”
Who is really leading the ground level application of technology?
A good example is Google DeepMind who was working with St Thomas' hospital to look at medical records and create algorithms. From my understanding, this was designed to help analyse the way information regarding patient records were processed and moved around the different departments in a hospital. In the future, this could different healthcare conditions and then point them in the direction of which is the best treatment for these people. DeepMind is widely regarded as the leaders of deep reinforcement learning. Their research in the creation of Neural Turing Machines and other programmes looking at replicating human short-term memory could effectively analyse patients data to make an assessment of what will probably happen to a particular patient in the next 5-10-15-20 years.
Another great example comes from a startup project in India, which is using AI to radicalize and automate the diagnostics. The guys behind it are machine learning graduates, and they have focused upon affordability to help doctors analyze people's blood results. They have taken machine learning and robotics and integrated them into things already used in laboratories — microscopes, assays, mobile phones, etc. These guys have created an ingenious solution that is low cost — it's effectively a rig of readily available equipment such as a smartphone and microscope. It takes photos of what the microscope is looking at and the microscope is being controlled by a basic robot control arm. The system analyzes what it sees in the photos via machine learning algorithm to help the doctor identify what the diagnosis may be. Similar to DeepMind's, which will be able to identify exactly what the cause is within 99% error margin, this system will be able to identify what the disease subset is and then automatically report back to the doctor. Such technologies greatly enhance the life of the multidisciplinary healthcare team — not “destroy” their role in society.
Such developments effectively reduce costs in data administration, medical records analysis, diagnostics, waste, facilities management and promote preventative healthcare measures — all the things that many CFOs and transformation managers will tell you to have to happen, especially in public healthcare systems. This is a huge problem around the world, especially in rural parts of Africa, Asia, South America etc. At the moment, the delay between a doctor taking a blood sample from a patient and then getting the results back can be a couple of hours. With such technologies, it'll be a couple of minutes. Machine Learning is most exciting in its scale — the sheer volume of people which can be analyzed, increases exponentially. Simultaneously, such a system could have analysed 5000+ patients within an hour, whereas 5 human beings could never do 5000 patients, as it is physically impossible. Think deeper into the future and Quantum technologies such as those being developed by D-Wave, 1QBit, IBM, Qubitl India or Rigetti could ramp up the processing power and hence scale to unknown levels.
But are we there yet? The answer is a definitive “no” — there are so many complex problems to solve, so many moral and ethical hurdles for humanity to consider. The participation and democratic oversight of the use of personal data has not been explored yet. Then, we must consider the wider social impacts of this technology. Is the path we are taking — as a human race — a good one? Will we actually reach artificial general intelligence in time to prevent the emerging disasters of climate change, political unrest and the financial austerity? I think the future is incredibly bright if we focus on organisations lead by academics endeavouring to help make people’s lives better.
About the Author: Amardeep Singh is a co-founder and CEO of an International Quantum technologies, software development and experimentation institution.