Submitted by Tim Mack on
Not all doctors view artificial intelligence (AI) with favor. However, instead of replacing medical personnel, using AI capabilities allows doctors and technicians to deliver medicine that is more personalized, proactive, and effective. This includes preventive medicine by proactively monitoring early warnings of illnesses, combining digital connectivity and AI analysis, and focusing on wellness in a cost-effective manner.
As a result, algorithms can monitor and adjust medication levels and combinations as a patient’s condition changes in real time. Accordingly, medical personnel can be more interactive than previously possible.
AI can be taught to review X-rays, MRIs, CT scans, and other diagnostic imaging that would otherwise await interpretation by a radiologist. This can include highlighting previously undiagnosed osteoporosis by scanning X-rays for hairline spinal fractures. Many doctors appreciate the radiology flagging, as it enhances their ability to quickly assess the images. And AI interpretation of mammogram results can now be done 30 times faster than by a medical doctor, with a 99% level of accuracy.
The increasing use of machine learning and language processing allows closer analysis of doctor–patient interviews. As a result, the doctor is not distracted by the need to capture notes in the exam room. However, some doctors have also been using shorthand reporters (like court reporters) to sit in exams and consultations, in the belief that overdependence on AI decisions is a path best avoided.
The Digital Trend
The dramatic increase in the ease (and the reduction in cost) of recording, storing, and analyzing data has been enabled in part by the fact that more than 90% of U.S. hospital records have been converted to digital format over the past decade. The use of AI in diagnosis, treatment options, drug development, and patient interface falls into this category and could reduce the average of two hours of digital and paper work that medical doctors must undertake for every one hour they spend with patients. In turn, what often seems like overwhelming levels of medical data could then be turned into actionable support for decisions.
One clear area is gene sequencing, where a previous substantial backlog (due to the complexity of the task) has been dramatically reduced over the past decades through machine learning enhancement.
Another baseline advance is more effective data management through improved AI algorithms for risk identification of such factors as falls and malnutrition to better determine which therapies might most benefit a specific patient.
Digital technology is rapidly improving the doctor–patient relationship. One use of virtual reality in pediatric care that has been growing since 2017 is patient headsets displaying scenes such as roller coaster or helicopter rides. Their use offers a distraction for those children affected by fear and high pain levels during injections or other unsettling or intimidating medical procedures. A study in Pain Management in 2018 showed 94% of juvenile patients surveyed reported a reduction of pain and fear, to the point of a failure to flinch during a previously terrifying shot.
Another example is RoboKind’s “Milo” robot, used to teach better social behaviors to children with autism spectrum disorder, such as the ability to make eye contact.
Education and Physical Therapy
There is growing use of AI virtual assistants to teach new languages or even organizational skill in a one-on-one format. But educators are concerned that this growth is also fraught with the danger of having the “Siri” analogs do the research for students, versus teaching them problem solving and research skills.
Algorithms from Bionik Laboratories Corp. in Toronto use a combination of AI analysis and robotics to assist patient recovery from stokes. Diagnosis of arm movement difficulty is combined with robot guidance of the most effective physical therapies. These techniques have now been found to outperform human therapists, in terms of effective productivity.
AI can be used to spot and flag suicide risk and other self harm through the scanning of social media (even Instagram photos) for evidence of depression and PTSD, for example, as well as suggesting optimal treatment for specific cases of depression. Such work is in development at Harvard’s McLean Hospital.
Drug Development and Use
AI is being used to improve trial design, drug study recruitment, and participation (especially for women and minorities, who have often been underrepresented in past drug trials).
Drug interactions—including those with over the counter substances such as nutritional supplements—can be found utilizing such resources as Semantic Scholar, with now more than 181 million scientific papers. This is just one example of specialized data banks that are expanding and refining their effectiveness.
Other AI Applications
- AI trained to detect early symptoms of macular degeneration in order to initiate earlier treatment.
- Algorithms developed by Eyenuk in Los Angeles to diagnose diabetic retinopathy (as well as an ophthalmologist) but in minutes.
- Predicting potential risk of death in heart patients in order to determine priorities for scheduling implantable devices.
While these examples only catch the high points of an amazing growth of AI potential for medical research and practice, it is clear that a revolution is underway. The more challenging arena is the cultural zeitgeist—how to ease those skeptical of the digital medical assistant into ease and comfort with their new colleagues and service providers.
“Artificial Intelligence & You” [cover theme] WebMD magazine, January/February 2020. https://img.webmd.com/dtmcms/live/webmd/consumer_assets/site_images/maga...
“Effect of virtual reality headset for pediatric fear and pain distraction during immunization” by Chad Rudnick, Emaan Sulaiman, and Jillian Orden, Pain Management, Volume 8, Issue 301 (May 2018). https://www.futuremedicine.com/doi/10.2217/pmt-2017-0040
Timothy C. Mack is managing principal of AAI Foresight Inc. and former president of the World Future Society (2004-2014). He may be reached at firstname.lastname@example.org.
Image by Gerd Altmann from Pixabay