Artificial intelligence has the potential to change the world, but nowhere is its influence being felt more than in healthcare.
Healthcare is undergoing a big data revolution. Health records are going digital and technology is allowing health services to collect more information than ever before. All this is about to facilitate another revolution, one whose scope could be incredibly far reaching: artificial intelligence. All that data is making it possible to design more sophisticated AI and machine learning technologies. This, alongside the proliferation of cloud computing, is opening up all sorts of opportunities.
Data in health services
Healthcare is a data driven world. Each case creates countless points of data which, if they can be captured, could increase the chances of the patient surviving. Unfortunately, much of this data often goes overlooked.
It was this concern, for example, which led to one doctor, Mohammad Al Ubaydli, creating a system which could put patients in control of their own data. The idea came about when he heard a story from a surgeon who had given a patient an all clear. The patient had taken his scans home and spent time looking them over.
When he spotted something which concerned him, he showed it to the doctor, who identified it as the early signs of a brain tumour. Such had been the doctor’s workload that he had missed it on his first examination, but because the patient had time to pay closer attention, he’d managed to spot the error, even if he didn’t immediately know what it meant.
The result is a solution called Patients Know Best which gives patients access to their medical records and allows them to share it with clinicians. It means that, when patients with long term conditions, are seen by a new doctor, they can quickly bring them up to speed with their medical history making it easier for that doctor to design effective treatment plans.
Authorities have also been looking to open up medical records. The NHS now akes it possible to access your medical records online and to have then changed if you think there’s an error. IT systems now makes it easier for health services to store and share data about patients between different services and departments.
As data becomes more mobile, it is leading to faster treatment times, better outcomes and a more pleasant experience for the patient. Artificial intelligence is the next logical step.
The AI revolution
AI is one of those technologies which has been talked about for many years. Its potential has long been clear, but what’s been lacking until now has been the ability to make it work. The big data revolution changes all that. Digital technology has created a surge in data which, if it can be captured and analysed, can deliver multiple valuable insights.
This is the fuel AI needs. To function properly, it requires vast amounts of information to design algorithms complex enough to create systems which can deliver intelligent and automated solutions.
The first applications have come in the form of relatively simple solutions such as chatbots, virtual nurses and administrative systems. For example, Babylon Health is an online medical consultation service. Using medical records and common medical knowledge, the app uses speech recognition to help users identify their symptoms. Users describe their ailments which the app then checks against a database of known diseases. If a patient needs more help, he or she can be referred to a doctor for a video consultation.
Care Angel, meanwhile, is a virtual nurse which extends care beyond a hospital setting. Using AI, the system can monitor patients in the home, check they are okay and take information. For example, the system can call a patient and ask how they are feeling. If they have a problem, they can take some medical information such as their blood pressure. If this is abnormal it can signal an alert to a healthcare professional which tells them a person needs to be seen. It is designed to give patients closer monitoring and to reduce the rates of readmissions.
Automated admin systems are also reducing paperwork. Healthcare has always been heavy on the administration with a lot of records and documentation which must be checked and stored. Digitisation of records enables automated systems to take care of many administrative tasks which might previously have taken hours of work. It frees up staff to concentrate on other areas such as patient care and can also reduce the risk of human error in record keeping.
Into the future
However, this only scratches the surface of what technology can achieve. The more data health services collect the easier it is to power increasingly sophisticated algorithms to deliver even more value. Indeed, Accenture believes that clinical health AI applications could lead to annual savings of $150bn by 2026.
More importantly, it can save countless lives by improving the diagnosis process and helping surgeons.
These technologies rely on the ability of AI to improve data analytics to capture both structured and unstructured data. Structured data comes in the forms of lists and tables which can be relatively easily stored and managed, but unstructured data is another issue entirely. This may take the form of images, videos or audio files which can’t so easily be placed in a single category.
In the healthcare sector much of the most useful data comes in the form of unstructured data such as X rays or MRI scans. It needs the application of AI to extract insights from all that rather complex information.
Deep learning can be applied to train algorithms to detect abnormalities in scans and pick up problems which might have been missed by human doctors. For example, brain scans can be analysed to detect signs of cancer or haematomas and increase the numbers of cases which are identified early.
In an environment in which a patient’s survival chances depend on how early their conditions are diagnosed this could increase survival rates dramatically.
Already, unstructured data is being put to use in the surgical setting. An AI platform called Caresyntaz, for example, aggregates both structured and unstructured data from the operating room, devices, electronic health records and other sources to present it in a unified dashboard which can inform clinical decision making. It reduces the amount of time surgeons spend on their cases and helps them to make better and more informed decisions.
VizAi, meanwhile, can view brain images to detect a stroke and send alerts to the attending physician’s smartphone. If signs can be detected early enough, strokes can be avoided in most cases with the right interventions. The system promises to reduce the heavy burden on health service budgets caused by stroke patients.
The exciting thing about AI technology is that it has the ability to deliver so much more than it already has. As machine learning comes of age, systems will be able to learn as they go, theoretically ensuring they become ever more effective. We’ll see it combining with other emerging technologies such as virtual and augmented reality to help in surgery, training, remote healthcare and other areas.
It is being combined with genome analysis to provide precision personalised medication. Algorithms can identify patterns and spot problems much earlier than would have been the case previously. It can develop a complete personalised risk profile based on a person’s genes and their lifestyle choices to develop treatment plans and help people to make certain changes to reduce their risks.
There are problems in the way. It is a fast moving and relatively hard to predict field. Developers of AI technology will promote themselves as being able to match or better the performance of healthcare professionals but this may not always be the case. There is also a danger of professionals becoming overly reliant on the technology in making their decisions. There is a question of how it should be used. Should it confirm a decision of a clinician or should it issue an alert first and guide the entire decision-making process?
Its reliance on data also heightens the vulnerability of the health service to cyber-attack. Patients are cautious about sharing their personal information with any organisation unless they know it will be safe. Health services have been slow in the past to keep their cyber security processes up to scratch which makes them prime targets for cyber criminals.
Challenges remain, but the potential is vast. We live in a world in which health services are struggling with budgets. The cost is growing all the time, and the pressure is on to deliver dramatic advances in outcomes and operational efficiency. AI is one of those rare technologies which can reduce costs and workloads while also delivering major improvements for patients.