How often when someone asks a question that we don’t know the answer to, do we immediately grab our phones and either ask Siri or look it up on Google? As a society, we no longer have to learn and memorize as many facts as in days gone by, which can be seen as both good and bad. Not to get entangled with an argument on this, but to look at the benefits within the healthcare realm, things like artificial intelligence and machine learning saves lives for a myriad of reasons.
Machine learning in healthcare is defined as a technological assistance for providers to identify and predict adverse incidents by learning from past and present situations. Self-learning computers and software is being utilized in numerous other industries very effectively, however, implementing it into medical settings is much more difficult because of the type of data that is being collected and analyzed. It is very subjective to look at an x-ray and compare it to another individual’s; the value of an x-ray doesn’t lend itself to helping every other patient that is being seen. The complexity and quantitative factors are being assessed, and over time it may very well be possible to measure healthcare data in a way that is beneficial to everyone.
What is being employed with machine learning is very much along the lines of the Google effect, where a doctor or healthcare professional doesn’t need to know every detail about treating a patient, but pulls up a patient’s healthcare record and couples that with a database-like system that looks at that individual, as well as everyone else that has had similar types of symptoms, and then can produce an analysis specific for treatment.
The analysis takes into consideration many different factors that are hard to fully comprehend without the help of a machine learning system. In a very methodical and unbiased fashion, a machine learning system looks at:
- The patient’s individual health history
- The patient’s recorded family health history
- Current statistics of the patient such as vitals and prescriptions
- Current and past lab results
- The patient’s demographics
- The patient’s living conditions
With this kind of information and connecting that with known population health issues, and correlating symptoms from previous patients, it is possible to generate a risk assessment, which in turn helps the medical providers to supply the best individual care for each person dependent upon their needs. For example: many hospitals are working on being able to predict patients that might develop a central line associated bloodstream infection (CLABSI). Within the algorithms, collected data and personal information, a model is created that helps to predict the likelihood that a CLABSI could occur and is accompanied by actionable information that physicians can work with.
Information like this, as well as many other predictive models makes it possible to not only save lives, but to make healthcare a more precise work, especially on a patient-by-patient circumstance. Machine learning isn’t solely about gathering information on patients and producing an action report, but goes to the ability for machines to learn patterns and trends without having to be programmed with the information. Currently, the learning on its own is still a work in progress, but that is where just about everything technologically based starts out. As more information is aggregated, and as more accurate models are shaped, you will see a growth in possibilities and uses for machine learning.
“The pace of progress in artificial intellegence (I’m not referring to narrow AI) is incredibly fast… you have no idea how fast it is growing.” Elon Musk
If you’re wondering if robots are going to take over the healthcare industry and all of your care will be handled by something out of a sci-fi movie, you have nothing to worry about. Healthcare is learning and benefiting from insight gleaned from drilling down and understanding all the data being generated, but there are situations where human influence and touch cannot be replaced. Judgement calls are a long way off from being exclusively being made by machines, but just like turning to Google, there is a plethora of information that we couldn’t even begin to memorize, but is at our finger tips. Machine learning saves lives every day, and the potential to save more lives and improve the overall care received is only in its infancy. What we have now will probably look very different in the near future, and may look completely different a generation from now.