After a slow start, it seems like AI is beginning to disrupt healthcare in earnest. Major players from healthcare and the tech industry are arranging deals, ramping up the data collection needed for AI algorithms to work and finding new ways to apply the technology to healthcare.
Companies use the technology in a variety of ways — providing significant benefits for hospital operations, but also raising concerns about the increasing collection and use of patient data that these AI algorithms depend on.
AI is making a significant impact on hospital operations. Here are four ways that experts expect it to change how hospitals run in 2020.
1. Advanced Medical Diagnostics
Medical organizations are using AI to improve the accuracy of diagnoses of diseases like lung and breast cancer, as gold-standard diagnostic methods continue to suffer from high rates of both false positives and negatives.
In one study from earlier this year, an AI algorithm developed by a team of Google researchers was more effective than radiologists at detecting cancer in mammograms. Another study — in which researchers used AI to look for evidence of lung cancer in CT scans — provided similar results, with the AI performing even better than radiologists.
In practice, these algorithms could enable quicker diagnoses, enabling earlier and more aggressive treatment of diseases like breast and lung cancer.
2. Pulling Better Data From Records
Providers often lack access to the comprehensive data they need to treat a patient adequately. It’s normal for the entirety of a patient’s medical record to be scattered across multiple documents — or for data to be stored, unstructured, in one report — making it difficult for providers to find the most relevant information.
Some healthcare experts hoped that Electronic Health Records (EHRs) would help combat this problem — so far, however, the new systems haven’t been quite successful, with most physicians finding them to be time-consuming and inefficient. Data can remain hard to find because it’s often entered in the form of lengthy clinician notes — a problem that’s made worse by content importation systems that can duplicate existing information.
Because information is in everyday language, standard computer algorithms can’t pull out and organize relevant data in a way that providers can refer to quickly.
Natural language processing (NLP) may change this. NLP algorithms can effectively read written language. Under ideal conditions, these codes can sort through natural language as quickly as a typical computer algorithm works through a spreadsheet to extract information.
An NLP algorithm working on an EHR can isolate relevant details, keywords and context — effectively summarizing a patient’s medical history without the need for a human reader to scan through the document and note the highlights.
A few different tech companies are already offering experimental versions of NLP for EHR — like Amazon, whose Amazon Comprehend Medical can quickly sort through unstructured medical text and offer providers the most relevant information.
3. AI in Medical Devices
AI hospital improvements will also enhance the effectiveness of medical devices.
For example, AI is used in surgical robots to stabilize the recorded hand movements of surgeons, removing hand tremors or jitters that could make incisions and robot movements less precise. New CT scanners are using AI to automatically reduce noise and stitch together scans, making them easier for radiologists to read and interpret.
While these devices have been slowed down by the lengthy FDA approval process in the past, a new set of FDA rules may make it easier for manufacturers to get their AI-powered medical devices approved, speeding up their development and adoption
4. Increased Security Concerns
AI’s adoption isn’t likely to be a universal positive. The increased use of the technology has raised concerns around both security and the ethical use of data.
In at least one of the major projects involving hospitals and big tech that is happening right now, Google’s Project Nightingale, neither patients nor doctors knew that their data was collected and used. While the project seems to be compliant with HIPAA, security experts expressed concerns that the tech giant didn’t gather the information ethically — and that nothing was stopping Google from using patient data in unrelated projects down the line.
As hospitals collect and store more health data, security vulnerabilities will also become a serious issue. Information is valuable — and whenever companies store data in significant amounts, hackers will try to gain access. The healthcare field has already seen a few major security breaches in the past few years, and experts believe that these breaches will grow more common in the near future.
Medical organizations that want to incorporate AI into hospital operations will need to understand the importance of good data stewardship, as well as the potential risks that come with holding on to vast stores of patient information.
AI Is Changing How Hospitals Are Run
New applications of AI and increased tech sector interest in healthcare is driving hospital adoption of the technology. Already, companies use it to improve diagnostic accuracy, make EHRs easier to use and build better medical devices.
However, AI may create new problems for hospitals, as well. Data security is likely to become a significant, and hospitals that adopt AI will need to understand the dangers that can come with handling large amounts of sensitive patient information.