This blog post was published on Hortonworks.com before the merger with Cloudera. Some links, resources, or references may no longer be accurate.
When Benjamin Franklin said, “An ounce of prevention is worth a pound of cure,” he was talking about fire safety. Nevertheless, the axiom works just as well when taken literally. In fact, Franklin’s advice anticipated hundreds of years of healthcare best practices.
Spotting and preventing medical problems early on is far cheaper and more efficient than catching them late. The problem for overworked physicians is that issues are not always easy for human eyes to detect. Artificial intelligence in healthcare is helping to fill in the gaps by diligently mining as much data as possible and making helpful suggestions that can lead to potential medical problems being identified earlier.
The Need for Artificial Intelligence in Healthcare
Staff shortages mean that healthcare workers need all the help they can get. With experienced baby boomer practitioners retiring, there are fewer care providers to go around. In 2015, 37 percent of nurses said that shortages had worsened in the last five years. By 2017, that number had grown to 48 percent. Physicians—especially residents—must often fill in the gaps by spending up to 80 hours each week in the hospital, often covering 28-hour shifts. Long working hours and abundant workplace stress create the perfect conditions for error.
Humans can detect patterns in data but it can be a tedious process at which machines are better suited, particularly when there are lots of variables or scenarios to reference. Combine this fact with the overwork and shortage of time doctors must contend with, and it becomes even easier for them to miss telltale signs that could affect a diagnosis. Artificial intelligence in healthcare can help by surfacing signals that well-meaning physicians may otherwise miss.
AI does this by processing vast amounts of historical data using statistical models that identify patterns in all kinds of data. It can then process new data against the historical model to look for similarities that may not be immediately evident to a human physician. Hospitals might experiment with this technology to scan medical images and evaluate possible diagnoses based on a large selection of similar pictures.
How AI Is Helping Patients
As an example, Microsoft is powering InnerEye with the same technology that was initially created for the Xbox’s Kinect peripheral. This CT scan analysis system is helping oncologists cut analysis times from hours to seconds.
But using AI can go beyond spotting existing health problems—it can even predict emerging ones. By analyzing a wide variety of data inputs, machine-learning algorithms can detect potential problems before a human could have known about them, helping to make treatment less expensive and more efficient.
Medical software company Clearsense is working to ingest a wide variety of data from healthcare monitoring systems. The company processes that data in real time through machine-learning algorithms to detect subtle patterns in patients’ medical conditions. The software can sound the alert for clinicians, warning them hours in advance that a patient’s health could deteriorate.
This predictive capability often manifests itself in unusual ways. At Silicon Valley’s El Camino Hospital, the rate of dangerous falls among patients fell 39 percent thanks to AI software that took data from real-time monitoring systems. Every time a patient activated a light or an alarm, the software combined this information with their existing health record data, which contained information about treatments they’d had. The algorithm was able to alert nursing staff that a patient was at high risk of falling.
Using AI in Other Areas of the Healthcare Industry
The benefits of AI are clearly visible at the “sharp end” of the healthcare industry, where physicians and patients meet. However, machine-learning algorithms also have a role at other points in the healthcare value chain, as they help hospitals make improvements behind the scenes.
In some hospitals, data about treatment times is being processed to help predict how long a particular surgical procedure will take, or how long an imaging device may be in use. This helps to avoid waste and improve patient outcomes by building treatment schedules that maximize the use of scarce human talent and technology resources. For example, healthcare services company Vizient is applying AI to help its members deliver cost-effective healthcare. Pattern recognition on lab and patient data can provide recommendations on how customers can improve healthcare outcomes efficiently.
Data is the key to these advances: today’s machine-learning systems cannot get enough of it. Consuming vast volumes of historical and real-time data leads to more accurate AI models and better preventative care. The challenge for hospitals lies in processing that data, which often arrives in high volume, thanks to high-resolution medical imaging. Healthcare providers must have access to lots of computing power and storage capacity as they turn this data into insights. They must also take a careful approach to governing that data. Healthcare information is some of the most sensitive in existence, so protecting patient privacy and security is a primary objective.
Looking to the Future of Artificial Intelligence in Healthcare
Tools that help to manage these large data volumes, and audit who has access to them, are already supporting hospitals as they reap the benefits of AI-assisted medical care. Creating the right platforms to manage medical data now will build a solid foundation for AI developments in the future.
That future is looking increasingly bright. As AI algorithms become more sophisticated, we are already seeing cognitive computing tools digesting reams of natural-language medical research, condensing it into a corpus of knowledge to rival that of any modern physician. Does this mean that the doctors of the future will all be digital? Of course not. Human understanding, intuition, and empathy will always have a place in healthcare. There is a “last mile” in patient care that no amount of software will traverse. Nevertheless, tomorrow’s doctors will be more informed, more confident, and far less stressed, thanks to a digital ecosystem that supports their decision-making processes and reduces the probability of error.
Read more about how predictive analytics and big data solutions are changing the healthcare landscape.