AI in the Healthcare System Right Now
In the near future, you may be relying upon a mobile app for help with medication or any health-related questions you have for your doctor right now. But what if we could use Artificial Intelligence (AI) to tell if you were having a cardiac arrest, or if you had cancer? AI is being used all the way from virtual nurses to diagnosing cancer with higher accuracy than doctors do. The scope in which AI can be used in the healthcare system right now is so large, it’s practically endless. Here are some of the application of artificial intelligence in healthcare right now.
Cancer Detection:
A program created by the Houston Methodist Research Institute in Texas can diagnose with 99% accuracy and 30% faster than a human doctor if the patient has breast cancer. The issue right now with cancer detection, and more specifically, breast cancer detection is that many yield incorrect results. According to the American Cancer Society, in the US 12.1 million mammograms (which are used to examine for breast cancer diagnosis and screening), yield incorrect results half the time. Why this occurs is due to a lot of mammograms being interpreted incorrectly and women being called back for another test. This leads to an unnecessary amount of retesting occurring and it's simply inefficient, as well as highly detrimental sociologically to the women which are called back.
Breast Cancer detection is also a highly inefficient process as it takes 2 clinicians manually looking over 50 charts, 50–70 hours to process them. What the software can do is predict those that are at risk of developing breast cancer which allows doctors to carefully monitor them, this is done through the software analyzing the mammogram and translates it into diagnostic information (30 times faster than a human doctor and with 99% accuracy). A lot of biopsies are conducted after they are told they are at risk of breast cancer, and 20% are conducted unnecessarily as they didn’t need it.
How the AI software works is that it takes patient charts and collects diagnostic features from it, most likely using image detection and machine learning. It then used the mammogram findings it collected to diagnose a breast cancer subtype, getting more efficient and accurate results than a doctor would.
Cardiac arrest:
A program called Corti, uses information it collects through an emergency phone call made to dispatchers to detect if someone is experiencing a cardiac arrest. It collects information from the description of the incident provided by the person in need of help, the background noises in the call, to the tone of the voice of the person speaking on the phone. Using voice recognition and algorithms which help detect patterns in the data gathered through a phone call it is able to detect if someone might be having a cardiac arrest. When tested in Copenhagen in 2014 over the course of 161,000 emergency calls, it was able to identify accurately 93% of the time if someone on the line was experiencing a cardiac arrest. Dispatchers were able to accurately identify the condition only 73% of the time.
Currently, the issue is that many people suffer from cardiac arrests and studies from Denmark have shown that “a patient’s 30-day survival rate triples when a dispatcher recognizes cardiac arrest during an emergency call.” If the dispatcher is able to recognize through the phone call that it is cardiac arrest they can give instruction on CPR or how to locate the nearest defibrillator.