Health Data

Getting to Know Artificial Intelligence and Machine Learning

This blog explores health informatics—a collaborative activity connecting people, process, and technologies to produce trusted data for better decision-making.

By Clarice P. Smith, RHIA, CHP

 

Artificial intelligence (AI) has begun to permeate our professional and personal lives, with the goal of improving quality of life. AI is providing some very exciting functionality to existing applications and new uses are being introduced to the market almost daily. So, what is AI and machine learning? According to the National Institute of Standards and Technology, “Artificial Intelligence (AI) refers to computer systems that think and act like humans, and think and act rationally.”1 Machine learning is an application within a system that uses information to learn without human intervention or programming. As with any advancement in technology, there are some concerns.

Examples of AI in our daily lives include:

  • Vacuuming robots that learn the layout of a room and vacuum based on the data
  • Apple’s “Siri,” which answers questions and provides information to iPhone users
  • Self-navigating cars
  • Personal assistants/smart home devices such as Amazon’s “Alexa”
While AI enhances our lives, it also raises significant privacy and security questions. For example, is the vacuuming robot sending data to a third party regarding the square footage of your home? Is it possible the self-navigating car could track every location the user stops? Users of personal assistant-style smart devices might wonder whether the device is recording (and sending) conversations that it shouldn’t be—especially after such an incident occurred with an Alexa device in Portland. Can a hacker gain access to your devices and monitor what happens in your home?

Loyalty cards are already being used by both online and brick-and-mortar merchants to assess purchasing patterns, track purchases, and make suggestions about restocking supplies. Online, merchants track what products users view and make suggestions for future purchases based on those views.

In our professional lives, many applications now have some form of AI/machine learning such as computerized coding systems for data analysis/management, voice recognition programs that improve capture with use, avatars for behavioral health group therapy, personal assistants that make appointments and adjust dates and times based on the patient’s availability, remote patient monitoring for chronic illnesses, imaging diagnostics, and more. AI allows users to process massive amounts of data and information that is impossible with traditional methods and applications. From a privacy and security perspective, there are the usual concerns. For example:

  • Is the data being stored by a third-party vendor and sold?
  • What are the security features and is the application, patient and organization protected from intrusion and breach?
  • Is the data being used to monitor usage?
  • What if the application makes a mistake in learning?
  • What process, policies and procedures are in place to validate the system and to ensure patients are not harmed or data is not released erroneously?
There are many other opportunities and concerns presented with the explosion of AI functionality in the marketplace. Individuals should research all relevant personal privacy and safety information before purchasing any product. Organizations must put into place processes, policies, and procedures when implementing AI to optimize the functionality while minimizing or eliminating the risk. The increasing use of AI presents exciting opportunities for health information and informatics professionals to provide input and expertise in addressing the challenges presented by this technology.
Note
  1. Russell, Stuart and Peter Norvig. Artificial Intelligence: A Modern Approach Third Edition. Pearson, 2009.
 

Clarice P. Smith is director, health information management department at AnMed Health.

Tags: health data , informatics , artificial intelligence , machine learning