There continues to be rapid development and take up of Artificial Intelligence (AI) solutions for healthcare delivery in diverse areas such as wellness, detection of disease, diagnosis and treatment, which is creating further transformational opportunities for advancements in private and public health. This article outlines some of the key considerations in the development of AI in the healthcare sector.
Myth or Reality
AI is part of what is now termed the 'Fourth Industrial Revolution' and is one of a number of more recent exponential technologies such as biotechnology, blockchain and 4D printing. It is not a new science but has been with us for decades and is developing rapidly, not least because of the increasing power of computers which has provided capacity for growth. This growth of computational power over a short period has been at the core of the AI profile and take up.
AI is defined as 'simulation of human intelligence processes by computer systems and other machines'.
This points to a form of copying or mimicking of the human to solve complex problems automatically through the power of technology and the computer. Segments of current AI include Artificial Narrow Intelligence which is quite focuses on a particular area i.e. research work such as Google Search or IBM Watson. Also, Artificial General Intelligence – this is broader and is AI that can respond to the environment like a human would with human cognitive abilities.
Much talk and media attention has surrounded the introduction of robots in hospitals and the home. How will they be controlled, who is responsible for them and are they safe? Whilst robotics for healthcare is very much a reality and the event heard of robots teaching themselves to walk and doing complex eye surgery, the focus is on improving our day to day operations and use of data with AI developments enhancing and improving the delivery of healthcare with more efficiency and accuracy to the patients with the advances embraced in doctor and clinicians daily work.
Healthcare is priority for governments across the GCC and globally, and the costs associated with investing in healthcare improving technologies is steep. The sector is a heavy investor in and user of technologies and it has vast amounts of data concerning patients and health that with appropriate capture, analysis and use can bring massive benefit to day to day provision of treatment as well as significant medical research and preventative solutions.
Lots of Data and Information – analysis and precision.
With all the information and data available to the healthcare networks in different locations and of a diverse and complex nature, if this can be easily and efficiently accessed and analysed by technology, it can provide speed of response and accuracy in identify health issues and solutions. AI can achieve and is achieving that. For doctors to keep up with all the latest medical research and practice, systems that will assist in making that more accessible in treatment will be key to optimal treatment of patients. Computers can be better able to identify, through imaging, particular conditions or problems that the busy doctor and radiographer might do alone.
Data is key – but it must be the right and accurate data
AI relies on data – often large amounts and complex data. The solutions from the mining of this substantial data can be significant for healthcare in research, diagnosis and treatment and day to day areas of wellness. Mobile solutions are now proliferating for the individual to use in daily living and monitoring connected to primary care facilities and hospitals to address warning calls and emergencies. However, the data must be accurate to ensure the analysis and delivery of solutions will be effective. This requires particular care in planning and preparation for delivery of AI healthcare solutions and measurement of the outcomes to confirm results.
Advancing Rapidly – evidence it is reality
It is not just 'market hype' and there are already some very significant and recent examples of AI developments and certainly more on the way.
We are seeing already diagnostics with 'body-on –the chip' solutions to test new drugs and side effects, automation of surgeries, technologies for DNA editing, abnormality detections for tuberculosis, MRI brain tumor and CT lung cancer detections – all examples of successful AI adoption. The scope for the full patient context to be available for examination and treatment is now enhanced with genomic profiling, imaging studies, family history, clinical and behavioural history and laboratory tests all available.
With Contextual awareness capabilities and advanced data analytics now readily available, AI is now delivering radiologists with the most relevant case-related information and tool sets to greatly assist them in their jobs.
Law, regulation and ethics
The advances in AI are so rapid and so 'disruptive' in potential that the law and regulators are being challenged to keep ahead. The technologies are also giving rise to ethical concerns and challenges.
Effective international efforts and cooperation will be required and it will be important to put in place regulatory standards. The UAE has been very proactive to this challenge with a new UAE Minister of AI appointed and a major international conference on AI planned for Abu Dhabi later in the year. This proactive leadership will be key to seizing the undoubted opportunities of AI but also assist in identifying risks and controls required. The law makers and regulators will also need to consider the legal frameworks and dynamism of the subject – this should be reflected by more principle based legal frameworks with the ability to respond more readily to innovations and threats and with provision to more easily implement more detailed executive regulations in support. It is a difficult balance to get the balance right between the sustainability of more dynamic and higher level principles and more detail so there is more clarity to address shorter term needs.
Existing legal issues and future considerations
With the DNA of AI around data, there are increasing requirements to consider data privacy and security. Healthcare information is a target for hackers and so capturing large amounts of valuable health data will continue to attract the unscrupulous.
The digitisation of medical records continues to be key in improving healthcare delivery but this requires robust law and controls. In the UAE there is not yet a specific data privacy law but there are principles of international and accepted standards of privacy in healthcare that can be followed and there are some local healthcare regulations that also provide specific requirements.
For example, Dubai Healthcare City already has its own Health Data Protection Regulations and there are other medical sector laws which provide protections.
The direction of travel is towards more law and regulation in this area and AI will need to be brought in with robust and up to date recognition of data privacy and security. A key plank of the new General Data Protection Regulations and regime which comes into force on 25 May 2018 and will directly impact the EU and many other countries, including the UAE, with connections to the EU, requires compliance with the principle of 'privacy by design' and this requires that data privacy and security must be considered as systems are built and not reactively. Businesses will not only need to consider privacy assessments but demonstrate they have carried them out.
Where businesses are looking to buy or sell AI related solutions it will be important to have in place commercial agreements that clarify responsibilities and liabilities and address areas including intellectual property rights, licensing and liability to third parties.
Robotics offer challenges over and above many that the health sector already faces. Duties of confidentiality that exist for doctors and healthcare clinicians are fundamental in law and regulation but where there are security issues this can test that framework. If there is outsourcing and use of third parties ie laboratories or research centres these can be outside the scope of the law.
And what of robots? The legal responsibility for and liability of the robot remains a challenging question. However, there are more obvious solutions for avoiding liability and regulatory attention with regard to AI solutions and robots, by having in place effective legal agreements with third parties that include provisions relating to liabilities, data privacy, security and other areas of legal risk.
The outcome of the AI revolution will undoubtedly bring much needed improvement to public sector and private healthcare programmes and potentially veryrapidly. This will impact organisations and how they embrace AI will need to be carefully planned for. It can certainly change working lives and the work environment so there are labour workforce that also need to be considered and worked through.
An exciting future – planning for change
The rapid development and take up of AI is an exciting advance in healthcare. The speed of change and the effect on daily lives and the traditional working environment brings with it challenges to organizations and the applicability of law and regulation – current laws can adapt and provide answers but in many areas the changes are hard to assess and so a new and flexible framework will also be required. In the meantime, organisations must plan carefully and follow existing laws and regulations ensuring commercial agreements manage potential risks and liabilities.