The development of artificial intelligence has always put people on edge. The one thought that always comes to people’s minds is that the robots are going to take over the world, and while the science fiction of it all is interesting to think about, if people take a minute to look at the reality, then they would see the extraordinary benefits it could have, along with the actual dangers that it can lead to.
Artificial Intelligence can impact all areas of life, but one of the main areas to be altered is healthcare. Even in the early stages of AI’s ability, healthcare is changing due to AI. For example, the FDA has approved a device, called IDx-DR, that detects diabetic retinopathy which causes vision loss. The FDA news release stated:
“IDx-Dr is the first device authorized for marketing that provides a screening decision without the need for a clinician to also interpret the image or results, which makes it usable by health care providers who may not normally be involved in eye care” (Office of the Commissioner, fda.gov).
Advancements within AI are becoming more prevalent, and changing healthcare in tremendous ways. The more work that is being put into development of AI programs, the greater the ability of the programs to run independently of doctors controlling or interpreting the results.
While there are multiple different types of AI programs, the one that shows the most promise is deep learning programs, in which the AI is feed different images, and processes them to begin to recognize patterns. Writer Paul Hsieh has seen these abilities in which, the “new “deep learning” artificial intelligence (AI) algorithms are showing promise in performing medical work which until recently was thought only capable of being done by human physicians” (Hsieh, forbes.com). The deep learning programs use the patterns it create in order to produce its results. As Dr. Robert Pearl, contributor to Forbes magazine, explains, in healthcare, the programs are fed scans in AI are capable of learning over time, and are able to differentiate between something that could, for example, be cancerous, or something that it benign. Unlike doctors these programs do not forget any information put into them and are able to diagnose in a fraction of a time than a doctor could.
The video above shows an example of the ability and benefits of AI diagnostics. It is of a TEDx talk by Dr. Shinjini Kundu, a physician and computer scientist. She speaks about how, in cartilage scans, the AI program was able to identify which patients were going to develop osteoarthritis and which were not three years before any doctors with a 82% accuracy. AI programs are able to predict years in advance, with more efficiency and accuracy than any doctor. With this kind of ability, AI should be trusted with making a correct diagnosis from imaging, and getting the patient a treatment plan faster than a doctor could.
Due to this exceptional ability of AI to identify different diseases from images makes it so that there can be more trust placed in a program coming up with diagnoses through scans. Though problem can arise for the future of medicine due to trusting AI blindly to make diagnoses, since a lot of medical discoveries come from finding the reasons for diagnoses occurring in a patient. Discovering why and investigating further into the reasons for a patient’s diagnosis can lead to new medical advancements, but without investigating into the why of a diagnosis, these advancements will come to a halt.
Just like every other kind of technology, AI it has its limitations, but the ability of the programs to diagnose patients has become more efficient and accurate. It has shown an amazing ability to diagnose patients, from finding tumors in a MRI, or diagnosing Alzheimer’s from a brain scan. These abilities are a huge importance to the medical world, with doctors needing less time to sit around trying to diagnose a patient and to being able to get to their treatment plan, and getting the patient healthy, faster.
With AI being able to recognize images so well, a lot more trust can be put into the diagnosis that comes out as a result. Doctors can use it more as a check for the final result, than another test on they way to a diagnosis. AI has even been seen to diagnose illness from scans years in advance over any doctor.
But just like most things in life, anything that could benefit the world, will always have a backlash if people are not careful. There are some actual dangers of trusting AI that people do not always think of when determining if it is a positive or negative step for the future of medicine.
One of the main problems that can come from integrating AI into healthcare is not putting in enough work into the studies before integrating the technology into clinics. Without doing enough testing of the programs, the AI might not work properly and can make more mistakes, and mistakes in healthcare can end someone’s life. As an article from Nature says, “Slow and careful research is a better approach. Backed by reliable data and robust methods, it may take longer, and will not churn out as many crowd-pleasing announcements. But it could prevent deaths and change lives” (AI Diagnostics Need Attention, nature.com). Just because AI is being integrated into healthcare business and being promoted does not mean that it is is ready to be trusted over a doctors opinion. Artificial intelligence being used in healthcare is an opportunity for companies to make more money and expand their business all at the same time. Companies may push AI as a way to spend less money on doctors and make money selling the AI. By letting AI have more control, this could lead to misdiagnosis and could harm a person in detrimental ways.
People have to also look out for becoming too reliant on AI. Humans can be inherently lazy, and with AI beginning to due their job, doctors may take it as they no longer need to look into what caused a patients disease, but only telling disease and treating it. Computer scientist, Sebastian Thrun, has pointed out that working with the neural networks “‘You cannot tell what they are picking up. They are like little black boxes whose inner workings are mysterious’” (Mukherjee, newyorker.com). AI is incapable of explaining how it comes to its conclusions, incapable of explaining why something is happening to a patient. Without the answer of “why”, doctors will not longer be able to establish ways for preventative care for future patients or explain why a current patient or their family why this has happened to them.
AI can have some truly amazing benefits, and can revolutionize how diagnosing works and the rate at which a patient can begin treatment. With early studies showing that AI diagnostics can be more accurate and more efficient, soon AI programs can be trusted explicitly for diagnosing a patient. But, people need to make sure that even if businesses say that AI can be trusted to make a correct diagnosis, that does not mean that enough studies have gone into making sure it is 100% reliable. Along with the wariness of that, doctors have a responsibility of not becoming completely reliant on AI, and need to keep investigating into the “why” of a diagnosis.
References:
“AI Diagnostics Need Attention.” Nature News, Nature Publishing Group, 13 Mar. 2018, http://www.nature.com/articles/d41586-018-03067-x.
“Artificial Intelligence Continues to Change Health Care.” U.S. News & World Report, U.S. News & World Report, 20 Sept. 2018, http://www.usnews.com/news/healthcare-of-tomorrow/articles/2018-09-20/artificial-intelligence-continues-to-change-health-care.
Hsieh, Paul. “AI In Medicine: Rise Of The Machines.” Forbes, Forbes Magazine, 30 Apr. 2017, http://www.forbes.com/sites/paulhsieh/2017/04/30/ai-in-medicine-rise-of-the-machines/#21ffb97fabb0.
Office of the Commissioner. “Press Announcements – FDA Permits Marketing of Artificial Intelligence-Based Device to Detect Certain Diabetes-Related Eye Problems.” U S Food and Drug Administration Home Page, Center for Drug Evaluation and Research, 11 Apr. 2018, http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm604357.htm.
Pearl, M.D. Robert. “Artificial Intelligence In Healthcare: Separating Reality From Hype.” Forbes, Forbes Magazine, 13 Mar. 2018, http://www.forbes.com/sites/robertpearl/2018/03/13/artificial-intelligence-in-healthcare/#e904bae1d750.
Ramesh, AN. C Kambhampati, JRT Monson, PJ Drew. “Artificial Intelligence in Medicine” The Royal Surgeons of England, 2004. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1964229/pdf/15333167.pdf .
TEDx Talks. “Artificial Intelligence Can Change the future of Medical Diagnosis | Shinjini Kundu | TEDxPittsburgh” YouTube, YouTube, 18 July 2017, http://www.youtube.com/watch?v=HrKzXLgGohA.
Walach, Elad. “Can AI Be Trusted Making Life and Death Decisions?” Forbes, Forbes Media LLC, 16 Feb. 2018. https://www.forbes.com/sites/forbestechcouncil/2018/02/16/can-ai-be-trusted-with-life-and-death-decisions/#fc277615951b
Images:
“The AI Industry Series: Top Healthcare AI Trends To Watch.” CB Insights Research, http://www.cbinsights.com/research/report/ai-trends-healthcare/.
“Diabetic Retinopathy | Pennachio Eye | Eye Care Eustis, Clermont.” Pennachio, pennachioeye.com/specialty-eye-care/diabetic-retinopathy/.
“Deep Learning: The Latest Trend in AI and ML.” Qubole, 3 Aug. 2018, http://www.qubole.com/blog/deep-learning-the-latest-trend-in-ai-and-ml/.
“Doctors Will Have Algorithms As An Extra Set Of Eyes To See Inside The Human Body.” GE Healthcare The Pulse, 12 Dec. 2016, newsroom.gehealthcare.com/doctors-algorithms-extra-set-eyes-human-body/.
AI-medicine 