Artificial intelligence is expected to transform nearly every industry in the coming decades, and the medical industry is no exception. We saw one early sign of that in August with the news that researchers from London’s Moorfields Eye Hospital and Google’s DeepMind subsidiary UCL had used deep learning to develop software that could use 3D scans to accurately identify dozens of common eye diseases.
The story was picked up widely, proving that AI in healthcare is of interest even if it’s also a field where the stakes are particularly high for technology that doesn’t work as advertised.
“AI is a hot topic right now, and its applications in healthcare, industry-wide, are exciting,” said Suvas Vajracharya, founder and CEO of Lightning Bolt Solutions.
Those applications for AI go beyond testing and treatments, touching everything in the industry from supply management to service delivery. “Much of medicine is a prime candidate for this technology, as it deals with diagnosis and treatment decision-making, which today is largely routine and leverages only small fractions of available information,” said David West, CEO of Proscia. “With AI, practitioners will be able to tap into data orders of magnitude larger than currently possible, in much less time.”
Here are some of the areas in the healthcare industry where AI is poised to change things in the coming years.
Clinical trials are one important area to watch for coming developments in AI and medical advancements, said Karim Damji, senior vice president of product management and marketing at Saama Technologies.
“The application of AI to natural language understanding engines is creating a revolutionary paradigm shift for drug development, catapulting the traditional user experience on an impersonalized dashboard to a conversational experience involving personalized analytics,” Damji said.
Saama provides a recent example of this, he said, through DaLIA, its deep learning intelligent assistant that uses natural language processing and natural language understanding to facilitate a conversational experience with clinical trial data that can make patient recruitment, process improvements, and analytics happen more efficiently and for less money.
“AI-informed solutions like DaLIA are the gateway to a new era of clinical development--one in which better drugs are brought to market faster than ever before,” Damji said.
Areas with highly repeatable tasks and high input/output information ratios are the ones where AI can first have an impact, West said. That includes molecular diagnostics, visual or image-based diagnostics (as seen in the eye research), healthcare operations, and drug discovery.
Operational processes in hospitals, clinics, pharmacies, and labs can also be improved and transformed via AI applications, said Mohan Giridharadas, the founder and CEO of LeanTaas.
“Our algorithms are able to ingest years of electronic health record data and apply data science and AI to learning how best to manage expensive constrained resources--infusion chairs, operating rooms, imaging equipment, in-patient beds and more--to improve patient access, decrease wait times and reduce healthcare delivery costs,” Giridharadas said.
Similar improvements in efficiency have been seen in other industries, such as the airline industry, through data analysis and service optimization, he said. Now AI and data science can be used in the healthcare industry to make similar improvements, helping patients have better experiences and making the healthcare system work more effectively.
Physician Health and Wellness
Patient wellness is, of course, an important consideration for the healthcare industry--but the health of the industry’s staffers matters, too, and it’s something Vajracharya has the potential to improve.
“Physicians and healthcare organizations are facing big challenges,” he said. “Physician burnout is increasing and the implications are costly, financially.” That burnout costs Stanford Medicine at least $7.75 million annually, for example.
“Tools that are going to help bring balance to medicine are essential,” he said. AI can be one of those tools, helping physicians work more reasonable hours by optimizing shift scheduling with data and measuring physician satisfaction to further improve the system.
The Next Level for AI
Another area of investing to watch is in the use of AI to analyze signals for diagnostics like a wearable medical device or genomic data, said Sean Ward, founder and CTO of Synthace.
“For example, Grail and Freenome are both using sequencing on blood samples to try to detect cancers before tumors or symptoms are visible, which is presently impossible for a human clinician to do at all,” Ward said.
“These will also be critical for the future of medicine but will require much larger investments and timelines to bring to market.”
Areas of Caution
“Domains with well-specified rules, such as games, can get around that by generating data to train on, but in the medical domain--where you may have to literally wait, sometimes years, for people to die before you can accurately label your training set--this is a serious limitation,” Verhaegen said.
“The availability of data places a definite ceiling on how far we can go with the traditional approach of trying to replace trained professionals with a piece of software.”