- Artificial intelligence (AI) in healthcare has the potential to dramatically change the industry. AI can impact how patients are treated and cared for and can affect cybersecurity and analytics. Although the technology is still in its infancy relative to other health IT infrastructure technology, many organizations are interested in AI.
While AI can be applied to many aspects of healthcare, all applications of AI involve adjustments to IT infrastructure. AI is another layer of technology added to the existing infrastructure that needs to be integrated correctly and monitored responsibly.
Organizations need to understand what the umbrella term of AI includes and what to look for in different infrastructure tools that include AI technology. Some tools may contain different AI methods to achieve a certain goal.
In this primer, HITInfrastructure.com breaks down the AI basics and highlights key benefits of the technology.
How is artificial intelligence used in healthcare?
AI was created to emulate the human mind and working processes, and can independently solve problems without needing to be programmed to do so. AI can accept new information and learn from it without human intervention.
The computing power behind AI allows it to process information exponentially faster than a human could, fixing problems or drawing conclusions that the human mind would never be able to achieve.
Gartner described AI as applications including autonomous vehicles, automatic speech recognition and generation, and detecting novel concepts and abstractions. Detecting concepts and abstractions is useful for detecting potential new risks and aiding humans to quickly understand very large bodies of ever-changing information.
The potential for AI in healthcare is vast and the technology can be applied from the infrastructure level all the way through treating patients.
For example, AI can be used effectively for cybersecurity. One of the biggest IT infrastructure security challenges facing organizations today stems from how broad their surface area is.
Connected devices are constantly being added to the network and entities are constantly expanding their cloud.
The wider surface area means there are more potentially vulnerable places cyberattackers can take advantage of. With more ground to cover, IT security staff can be stretched thin and legacy network security systems might not be able to catch evolving security attacks.
An ABI research report released late last year predicted that cyberattacks will do over $1 trillion in damages by the end of 2018. Cybersecurity vendors are considering AI to provide organizations with more dynamic and intuitive defenses to prevent these kinds of damages.
"We are in the midst of an artificial intelligence security revolution," ABI Research Industry Analyst Dimitrios Pavlakis said in a statement. "This will drive machine learning solutions to soon emerge as the new norm beyond Security Information and Event Management (SIEM), and ultimately displace a large portion of traditional AV, heuristics, and signature-based systems within the next five years."
"This radical transformation is already underway and is occurring as a response to the increasingly menacing nature of unknown threats and multiplicity of threat agents," Pavlakis concluded.
Analytics is another example of a patient facing use of AI in healthcare, especially when it comes to using images for diagnostics.
A computer with AI can look at an image of a healthy brain scan and an image of a brain scan with tumors. The device could then recognize the difference between the two images by breaking them down into machine-readable patterns.
The machine can remember and reference these patterns then apply them to future images to determine which patterns indicate that a brain tumor is present.
Machine learning and natural language processing
Machine learning and natural language processing are two AI methods used in healthcare. Machine learning is the method used mostly in image scans and natural language processing is used for text.
AI applications use machine learning to sort through the data collected by an organization to learn patterns and construct algorithms.
For healthcare organizations currently implementing or considering predictive analytics, machine learning can assist doctors and clinicians in detecting cancer and other chronic diseases early, based on patterns taken from scans of previous patients.
The subtle patterns that likely indicate the beginnings of a chronic illness can use previous positive and negative identifications to adjust the criteria for flagging future images from patients who have not yet received a diagnosis.
Natural language processing is mostly applied when text is involved and is particularly helpful when it comes to notes and EHRs. Many EHRs have the option for clinicians to enter free text notes on patients. This text is unstructured data, meaning that traditional analytics tools can’t process it.
Natural language processing algorithms are implemented to process the free text in EHRs so it can be referenced correctly. Natural language processing can tell the context of word that have multiple meanings.
For example, the tool knows that when a clinician types “cold” whether, she means the temperature or the illness, based on the other words in the sentence.
These two methods give organizations insight into their data that could not be achieved before because of the vastness of the data and its details. Making this data actionable will help improve patient care significantly.
AI’s infrastructure requirements
AI functions on collected data: The more data AI solutions have access to, the more successful its implementation will be.
AI solutions with a wide range of data are capable of making more connections. Additionally, these solutions become increasingly accurate in detecting, recognizing, and flagging concerning healthcare issues early.
The healthcare AI movement is largely connected with the implementation of the Internet of Things (IoT). Patients can potentially contribute health data to the network via wearable or other personal devices. These contributions to AI machine learning give the AI solution more data to reference.
The massive amounts of data and the increase in connected devices calls for a serious evaluation and plan for how an organization’s IT infrastructure can support the increases in activity.
Increased network security for IoT devices and advanced network security for organizations sending and receiving data with other organizations is necessary to ensure that the increase in devices accessing the network will be protected from malware.
Storage is another concern for AI implementation. Organizations will need to have at least a hybrid cloud environment to store data to increased data demands.
On-premise servers are costly and limited to a finite amount of storage space. The AI solutions needs to have access to cloud data constantly to implement machine learning.
Healthcare organizations are still several years away from fully realizing and benefiting from AI. Official guidelines and regulations are still needed to ensure protected health data remains secure and AI solutions are thoroughly tested before implementation.