Written by: Steven Tolle, Vice President, Global Strategy and Business Development, IBM Watson Health
We hear time and time again how Artificial Intelligence (AI) is needed in healthcare, but sometimes what gets missed is the why AI is so important to the future of health. Due to the explosion of health information in the form of patient records, breakthroughs in genomics, population health data and a steady stream of new studies and journal articles, healthcare providers are drowning in data.
In fact, medical data is expected to double every 73 days by 2020, creating a critical need for technology that can make sense of all this data and make it useful to help physicians deliver remarkable outcomes.
In radiology specifically, image volume continues to grow rapidly. In the past, imaging studies produced a handful of images, but advanced imaging technologies today produce hundreds or even thousands of images per study. A 2016 estimate done by IBM Watson Health concluded that in order to keep up with current imaging rates, radiologists would need to view an image about every two seconds — every weekday, all day long — in order to review all relevant images.
AI is ideally suited to meet healthcare challenges, particularly for medical imaging, in three ways:
AI thrives on an overabundance of data.
Much of the data in the electronic health records (EHR) is unstructured, such as physician notes or radiology reports, which is difficult and time-consuming to search through. As a result, providers’ time is being consumed by the hunt for relevant patient data, which impacts both workflow efficiency and patient care.
This massive amount of data makes healthcare a great place for AI to prove its value. The same image and data overload that is a burden for humans can help a deep learning system thrive. Through its ability to tackle large amounts of health data, AI systems can help provide quick access to a comprehensive patient record or the latest treatment guidelines.
AI can “see” what humans may miss.
In addition to tackling large amounts of data, AI is also adept at a function that is inherently challenging for humans: Seeing patterns that are outside the scope of where their attention is focused. In other words, people are good at finding what they are looking for, but not so good at finding what they’re not looking for. This human tendency is known as inattentional blindness.
AI systems, however, have no preconceived assumptions about expected findings that could blind them to unexpected results. Since missed findings can lead to negative health outcomes, the ability of a system to catch an abnormality that a practitioner may have missed makes it very valuable in healthcare — even essential.
AI integrates with — and helps leverage — existing systems and workflows.
AI systems are not very useful if they require separate infrastructure, systems, workstations or logins. An AI application that complicates workflows may make it challenging for organizations to demonstrate ROI, or may deter them from implementing it in the first place.
A well-designed AI solution not only is compatible with existing infrastructure and workflows — causing a minimal need for extra resources or disruption — but can help organizations utilize their existing systems more effectively. For example, an AI solution that can search for relevant patient information in the unstructured data of the EHR, such as IBM Watson Imaging Patient Synopsis, actually increases the value of the EHR.
It is our premise that radiologists are very good at finding what they are looking for but not at finding what they aren’t looking for. Radiologists are incredibly proficient at spotting lung nodules. Industry-wide, the miss rate for radiologists looking for specific abnormalities is just 3-5 percent. But that doesn’t mean they are seeing everything.
The near-term potential of AI is to build a safety net that lets us identify the high-value signals that might otherwise not have been the focus. Longer-term, the technology has the potential to revolutionize precision medicine and improve patient care. A lot still needs to happen before that long-term promise is fulfilled. But many of the critical building blocks are already in place today.
For example, we are currently able to use natural language processing technology to read clinical text from EHRs and progress notes to identify, categorize, and code unstructured data and turn it into actionable, quantifiable insights on a patient chart. That data is allowing us to highlight potential discrepancies in documentation and provide valuable clinical context to physicians during image interpretation.
This is a critical first step. Researchers from the Medical College of Wisconsin recently found that when radiologists had the time and access to patient charts, they would change their findings between 20 and 25 percent of the time. There is data locked in the patient’s chart that can be critical to a diagnosis.
The next step – which is being tested with radiologists around the world today – is leveraging those text analytics to inform care decisions. Initially, our work here was focused on specific organs, but it is evolving quickly to address specific conditions within entire body systems. What that means is that our technology will soon be able to screen chest X-rays and chest CT scans to help clinicians identify conditions such as emphysema and COPD, aneurysm, pulmonary embolism, pneumonia, and others.
With the capability to ingest large amounts a data, “see” hidden findings and fit into existing workflows, AI has great potential to help healthcare organizations achieve their central aim: improving their quality of care. By carefully nurturing this technology, partnering with healthcare providers around the world to train and test it, and aiming for consistent improvements in workflow processes, we are putting the pieces in place that will enable a real, sustainable revolution in healthcare.