REALITY Check: Artificial Intelligence

Feb 4, 2020

From interpreting scans to scanning hospital rooms for hazards, AI innovations are poised to change the healthcare landscape. Here are five ways the technology is being tested:

 

  1. Radiologists Phone a Friend

Radiologists at the NYU School of Medicine are looking for ways to cut down on unnecessary mammography follow-up screenings, a common occurrence which puts stress on patients and the medical system. They’ve developed a network that was trained using example scans and corresponding diagnoses and can now offer diagnostic advice. So far in testing the AI has shown modest but promising improvements in diagnostic accuracy.

 

  1. Monitor Glucose—No Blood Involved

For diabetes patients and others who monitor blood glucose, pricking a finger with a needle a few times a day is standard protocol. That could all change with new AI technology that detects hypoglycemia using electrocardiogram signals obtained through a wearable device. With an 82% accuracy rate, it’s as good an option as traditional glucose monitors—with real-time response and none of the hassle.

 

  1. The Startup for Better Bedside Care

Imagine if every hospital room had a designated nurse present 24/7. That’s what iN, a wall-mounted hospital AI device hopes to deliver. The device uses sensors to measure sound, light and temperature, and can even alert a nurse if a patient is at risk of falling. The device has been in testing for 8 months, and the creators have already seen promising preliminary results including increases in bedside reporting and decreases in falls.

 

  1. A Crystal Ball for Stroke Patients

A recent study by researchers at Taipei Medical University showed that AI models can be trained to predict outcomes for patients being treated with thrombolysis for acute ischemic stroke. Using prediction models, the researchers could predict major neurological improvement at 24 hours and three months following treatment. The AI model took factors like blood pressure and other measurements into account when making its predictions.

 

  1. Doctor AI Will See Your Results Now

Due to the prevalence of intense follow-up screenings and inaccurate diagnoses, research is underway to improve breast cancer detection. A new tool funded by Google Health aims to improve mistakes in both under- and over-diagnosis. After tens of thousands of scans were analyzed by the AI algorithm, both false positives and false negatives were reduced. The implications of this technology for radiology in particular are promising, as increasing burnout means that the volume of medical images is in danger of overwhelming physicians who analyze them.

 

These are just a few of the recent innovative ways that AI is being used to transform healthcare. As AI develops and becomes more universal, it could change the ways in which healthcare and technology intersect.