AI and medicine…where to begin? There is so much going on in how AI is impacting healthcare. It’s a meta trend that’s been developing over the last 20 years. I’m going to highlight 5 areas where AI is driving big changes.
The first technology is the brain computer interface or BCI, which you’ve heard me discuss before in my episode on brain hacking. BCIs are direct connections between computers and the human brain being used to restore function for patients who’ve lost the ability to speak or move or interact with their surroundings. This includes people suffering from ALS, strokes, or the 500,000 victims who have spinal cord injuries every year.
The second disruptive technology is AI radiology tools. Deep learning use neural networks, remember my discussion of them in my episode on deep learning and machine learning, Neural networks work in a way inspired by how the human brain processes information through many layers. Computers using neural networks have already proved their ability to match or exceed the accuracy of human experts when analyzing images. For example there’s a FDA approved AI program embedded in a mobile x-ray machine to identify and prioritize collapsed lungs on STAT x-rays. About 60% of all x-rays in a hospital are marked STAT. This AI enhanced x-ray machine flags images with possible pneumothorax (which is a collapsed lung) so those x-rays get looked at by a radiologist first, speeding up diagnosis and getting care to patients who are most ill.
Similar machine learning systems are used for more accurate detection of diseases from all types of imaging studies like CAT scans, MRIs, mammograms and even everyday detection of broken bones on x-rays.
A third impact is AI is being used in some cases to drive down to the pixel level of tissue biopsies seen under the microscope, thereby detecting changes not routinely observed by the doctors reading them. This is called digital pathology and is important because 70% of all decisions in healthcare are based a pathology result. The more accurate the image, the faster the right diagnosis is made and treatment can begin.
The 4th innovation is harnessing the power of smart phones using their great camera quality. Smart phone photos are analyzed by AI algorithms to diagnose skin cancer and eye diseases. And there are many other phone uses. So far there is a disposable sensor that plugs into a smart phone and can diagnose HIV, a glass ball that attaches to the smart phone camera that turns it into a microscope to detect malaria, and two clinical trials underway. One trial is testing a smart phone app that can diagnose acute respiratory problems in children by analyzing their cough and another trial is investigating the use of FitBits to collect data to diagnose Parkinson’s disease.
The fifth area is using AI to get ahead of chronic disease and could be where AI makes the biggest impact in the healthcare system. Clinically validated machine learning algorithms are being used to generate a patient’s risk for congestive heart failure, macular degeneration, aortic aneurysm and even hospital readmission based on medical data from their charts. Knowing which patients are at risk can lead to earlier interventions and even changes in their current treatment. In this way, AI generated clinical decisions using lots of patient data can make doctors more aware of the nuances in a patient’s health to get ahead of any developing medical problems.
These are 5 of the many, many ongoing impacts AI is making in healthcare. As a doctor it’s overwhelming to me just the medical applications of AI, and makes me even more committed to discuss AI in way everyone can understand, in a way that’s short and sweet. I’m Dr. Peper.