Artificial intelligence (AI) is set to transform the healthcare industry, poised to impact the cost, quality, and access to healthcare worldwide. From streamlining the drug research process, enabling personalized patient care, to fixing inefficient miscommunication within medical institutions, the list of ways in which AI will shape how we deliver patient care continues to grow. In fact, according to a research report by Global Market Insights, the global ‘Healthcare Artificial Intelligence Market’ is expected to surpass USD13 billion by 2025. The healthcare industry however, continues to present challenges that threaten to slow down this progress. What are the key opportunities and obstacles facing companies and institutions developing healthcare artificial intelligence, and how will they affect you?
We recently spoke to Susanne Suter, software engineer at Supercomputing Systems AG, at the ‘Women in Digital Health’ event where she was a guest speaker. She shared with us about these opportunities and challenges, and the innovative healthcare projects her team is currently working on to harness AI technology for good.
How do AI and machine learning work?
“When we talk about AI nowadays, we mean data-driven systems and machine learning,” explained Susanne. During a training process, machines are essentially fed a tremendous volume of high-quality data, what each piece of data means. When this learning phase is completed, the machines are able to predict or classify any new data that is inputted based on its stored knowledge – hence developing its intelligence.
This ability of AI to analyze tremendous volumes of medical data and recognize meaningful patterns within these data sets has proven groundbreaking in healthcare innovation.
The benefits of AI
For example, AI technology has made it possible to detect potentially life-threatening medical issues in patients early enough so that they can be treated quickly. Supercomputing Systems AG is working on a project with Prof.Dr.med. Emanuela Keller from the University Hospital in Zurich, to develop an AI research system that monitors and analyses real-time medical data of patients at the neuro-intensive care unit, predicting for example the occurrence of brain hypoxia in patients before it happens so that effective preventive action can be taken.
Pharmaceutical companies are also exploring ways to adopt AI technology in drug development; speeding up the drug discovery process and reducing costs. For example, Berg Health is using massive volumes of data from patients with diseases such as prostate cancer in order to identify new targets and develop new drugs. In Europe, global biopharmaceutical company Sanofi recently signed a 250 million Euros collaboration-deal with leading British drug design company Exscientia to discover bispecific small-molecule drugs against metabolic diseases.
The success of AI technology in healthcare, therefore, hinges fundamentally on its access to quality medical data – and lots of it. Collecting quality medical data however continues to present a real challenge for many companies working on harnessing AI technology for good.
Powering AI technology is challenging
Collecting quality medical data is time-consuming. “From my experience, 30 to 50% of time needed to develop an AI system is spent collecting quality data,” shared Swiss Supercomputing AG software engineer Susanne Suter. Furthermore, medical data in most countries continue to be siloed in disconnected systems that are difficult to access. In some cases, this data still resides in handwritten records stored in file cabinets in the doctor’s office. As expressed in Forbes, “Data must flow freely through AI systems to achieve real results but extracting data from handwritten patient files or PDFs is cumbersome for us, and difficult for AI.”
Successful collaboration by experts from fields including medicine and data science is also necessary in developing useful AI systems in healthcare. In a project with Dr. med. Peter Maloca from the Institute of Molecular and Clinical Ophthalmology Basel (IOB), Susanne’s team is developing automated analysis systems that can detect retinal tissues and medical conditions such as tumors in eyes. The team used over 2,000 images taken from an estimated sample of 650 different eyes. Each image had to be marked by experts, i.e. the retinal tissues are drawn on each image, before this data was fed to AI systems as part of the machine-learning process.
Additionally, companies developing AI solutions for healthcare have to actively consider issues related to cybersecurity and data protection. For AI systems that need to be connected to the internet to make use of powerful cloud-based backends, sufficient cybersecurity needs to be put in place to protect it against hacking. Because AI systems depend on the integrity of large medical data sets to be effective, sufficient measures also need to be put in place to ensure that medical data (e.g. from databases maintained by hospitals and other medical institutions) remain protected.
The potential AI in technology has in transforming how we receive medical care continues to grow. However, much of this progress still depends on the speed at which the healthcare industry successfully undergoes digital transformation. Additionally, as its success hinges on its access to a high quantity of medical data, questions remain about how the development of AI technology in healthcare will benefit from current efforts to return data ownership to the people – efforts that thereby free medical data previously siloed in medical institutions and restricted by data consent laws, back in the control of people who can consent their transaction.
“AI will ultimately help us in decision-making, but we are still quite far from a reality where machines can think and draw their own conclusions. And in thinking about how to collect quality data, it’s important to remember, the person who owns the data has the power,” concluded Susanne.
The future of AI in healthcare looks bright, and it will surely be an exciting area of healthcare innovation to watch over the next few years.
Susanne Suter, Dr. sc. Computer Science University of Zurich, has been successfully involved for over 15 years in multidisciplinary innovative projects at the interface between computer science, biology and medicine (including scientific prizes, third-party funds and publications). Since four years, she is working for Super Computing Systems as a software project leader and engineer producing custom-tailored medical software systems such as a patient monitoring system at a neuro-intensive care unit, second-opinion case-review systems for medical doctors, and an automated surveillance service to track the health condition in human eyes.
About the author
Aisha Schnellmann is a Singaporean sociologist by training, interested in healthcare, education, and sustainability issues. She is passionate about producing content that promotes meaningful dialogue, focusing on print and digital content that resonates with a strong call-to-action. Based in Zurich, her interest in digital healthcare grew from the conversations she had with committed medical staff in rural hospitals in Asia, who remain hard-pressed with the technology available to them.