What AI means for healthcare in 2019
Artificial intelligence (AI) is gaining significant momentum in healthcare, and the power of machine learning is forecast to have the potential to significantly alter the industry in 2019 and beyond.
Reflecting on the past 12 months
In 2018, a report released by PWC, Top health industry issues of 2018: a year of resilience amid uncertainty revealed AI is already gaining traction in healthcare’s back offices and supply chains, generating quiet efficiencies that perhaps aren’t garnering the same glamorous headlines as robotics and genomics.
According to the report, more health businesses around the world are now leveraging the power of AI to automate decision-making, create financial and administrative efficiencies, automate parts of their supply chains, or streamline regulatory compliance functions.
Repetitive tasks in particular may benefit from the introduction of AI and machine learning to replace or supplement human interaction, the report found. AI doesn’t forget, tire, get bored with tasks or develop carpal tunnel syndrome.
One of the ways AI and deep learning is already being used in healthcare is image classification, which enables extracting information from multiple images to help healthcare providers like radiologists mark file and mark low priority X-rays, making the process quicker, easier and more accurate.
Meanwhile in the UK, researchers at Oxford Hospital are using this technology to help improve diagnosis for heart disease and lung cancer. AI can also help primary doctors fund and refer patients to specialists faster, and offer more fast, accurate and actionable insights for doctors and their patients.
Freeing up time to focus on patient care
It’s important to note this doesn’t make humans redundant, in fact, it frees up healthcare professionals’ time to do less mundane tasks and focus on more personalised patient care. According to the PWC report, healthcare providers can leverage AI tools to help their staff analyse routine pathology or radiology results more quickly and accurately, allowing them to see more patients and realise greater revenues.
According to MedicalDirector’s CEO, Matthew Bardsley, increased education in the healthcare sector about ways in which patients and practitioners can better leverage technology, automation and wearables to optimise and share wellness data, can open up a fresh wave of opportunities to enable more ideal healthcare and a more patient-centric approach.
“The future looks promising,” he said. “The digitally enabled practitioner will be able to see their next patient, well-equipped with the same wealth of data that the patient has on their own wellness apps and devices – and more. The clinical visit will be more open, accurate and efficient, while the patient and practitioner relationship will become more trusting, personalised and transparent.”
AI and the age of wearable wellness
Leveraging data and insights from ‘wellness wearables’ can also aid diagnosis and treatment in healthcare. In fact, a recent study revealed wearable devices like Apple Watch and FitBit are already able to gather sophisticate data to enable serious detection of health conditions such as hypertension and sleep apnea.
“The age of wellness wearables is definitely here,” GP and MedicalDirector’s Chief Clinical Advisor, Dr Charlotte Middelton, said. “Whether it is the middle ear devices that monitor your heart rate, the wearables on your wrist that tell you how you’re sleeping, or the ECG monitors that go around your chest – there are so many exciting technology concepts that can really enhance the care we provide as doctors.”
Implications for pharmacy and clinical decision support
The PWC report also stressed that the implications of AI also extends to the pharmaceutical, drug and medical R&D sectors. For instance in the pharmaceutical industry, a company could use AI to automate the intake, analysis, follow-up and reporting of adverse event reports associated with their drugs.
When it comes to better clinical decision support (CDS), new inroads in semantic interoperability set to help reduce the length of time it takes from the moment new clinical knowledge is released, to the time it is practically applied in a healthcare setting.
Already, a study by the American Journal of Health System Pharmacy, Improving medication-related clinical decision support, has revealed how CDS systems should incorporate more patient-specific information into decision-making algorithms and employ human factors design principles.
Importantly, the report called for healthcare teams to be more accountable for improving interoperability, and ensure more regular updates of CDS systems to optimise accurate information sharing.
Taking this a step further, “Semantic” interoperability provides interoperability at the highest level, enabling two or more systems or elements to exchange information and to use the information that has been exchanged.
Applied to healthcare, this level of interoperability has the potential to support the electronic exchange of patient health information and data via health digital ecosystems, with the aim to improve quality, safety, efficiency, and efficacy of healthcare delivery.
So imagine a future where we have a digital ecosystem of semantic interoperability. Where we have AI-enabled CDS systems can analyse a patient’s characteristics and provide tailored recommendations for diagnosis, treatment, patient education, adequate follow-up, and timely monitoring of disease indicators – all at a fraction of the time and with more accuracy than ever before.
Time to get serious about data integrity
In order to best adopt AI as a true enabler in healthcare, the PWC report stressed the tool is only as good as the data it uses for decision-making.
Data integrity, or ‘data quality,’ refers to the process of maintaining the accuracy, reliability and consistency of data over its entire ‘life-cycle.’ Applied to healthcare, this can include (but is not limited to) maintaining the accuracy of patient’s personal details, health summary, clinical notes, test results and family information.
This is why the industry should start getting serious about investing in finding, acquiring and creating good data, standardising it and checking it for errors, the report said.
At the same time, healthcare providers should consider how their systems capture, collect, clean, integrate, enrich, store and analyse data. They should collect data in a way that allows it to be integrated with other relevant systems and in a way that can ultimately enable better healthcare now and in the future.
Respecting AI in a fear-based economy
Disruptive technologies, AI and machine learning can be powerful tools to drive better health outcomes. But in a fear-based economy, the AI-enabled predictions need to be fully understood within the context of each individual person and their circumstances, MedicalDirector’s CEO, Matthew Bardsley said.
“Imagine if an AI-enabled health prediction tool said to you, ‘you’re going to die when you’re 70.’ You then base all your life decisions based on that prediction, but come your 70th birthday, you’re still around, but you’re alone and broke, with no set future plans in place,” he explained.
The impact prediction has in a fear-based economy is very real and has a far more implications for our individual decisions, he stressed. If the fear doesn’t manifest itself or the fear is greater than what was initially predicted, these are much harder for the human mind to cope and take on board.
“So it has to be managed a lot more carefully, and it is effectively managing predictions in a fear-based economy that we need to bear in mind as we innovate in healthcare,” he added.
For AI to really be a powerful tool in healthcare and enabler of better health outcomes, Bardsley recommended we need to make sure it is done respectfully, there is a level of empathy in the system, and a layer of emotional artificial intelligence, before it can be unleashed into the healthcare ecosystem.
“Otherwise just one health prediction delivered incorrectly can lead to a life or death situation for a patient, and stifle innovation,” he said. “Moving forward, innovators in healthcare need to understand and respect this layer of emotional AI required, and really lean into this problem and understand its repercussions, so healthcare can enjoy the same capabilities as other parts of the market.”