Misdiagnosis of a medical
condition or illness is a huge problem particularly as it can potentially lead
to tragedy in life. Though doctors may seem confused by the fact that they can
misdiagnose their patient's symptoms, this is where the use of artificial
intelligence can assist them and the medical industry. Powered by a neural
network, the algorithm uses data collected from patient visits to try to
hypothesize what a patient is going through. By comparing data to the patient's
symptoms, the system generates solutions based on recognizable patterns
analyzed by the database. The author of this article even suggests that the
software was “more than 90 percent accurate at diagnosing asthma”, for example,
whereas, “human physicians who studied in that particular field were only 80 to
94 percent accurate” with their diagnosis.
Based off these statistics, it is proven that the software has accuracy to it.
However, will this be an issue to those who do not consent to provide their
personal information? In the United States the privacy policy of sharing
medical information is very restrictive; this can be prohibitive in obtaining
information needed to create a database with this purpose. As we know, medical
records contain very personal health-related information which some patients do
not want exposed. Though as the
article references, it is easier for China to succeed with this development of
A.I. technology, as their healthcare policy is looser with less regulations;
additionally, their population is enormous enabling them to have a larger
sampling. By having huge amounts of data being shared and collected with minor
restrictions, enables China access to more information to work with and the
ability to lead in the development of this A.I. technology.
Though only recently this
discovery has been made, will the future of diagnosing humans be replaced by
this software? We know that jobs in certain workplaces are being replaced by
technology, but will this technology also replace one of the most important
human jobs….a doctor? Quite frankly, even though the numbers prove the A.I.
worth, I am not sure how comfortable/reassured I would feel having only a
device tell me what is going on inside my body. For centuries we have relied on
a doctor’s medical expertise to reassure us of our next steps, but will our
trust be impacted by this technology? Like any research or data, manipulations can be made by
other people. While this information may be on record, that is not to say
someone like a hacker can go in and edit specific details. Though there are
many positive attributes to this software, technology has its risks.
Another great point brought up
in the article is that patients in countries, like India and China would
advance from employing this type of technology where physicians are scarce.
However, more testing is needed to determine its accuracy. Though this is a
long way into the future, action must be taken now to try to bridge the gap
between misdiagnosis and artificial intelligence accuracy.
Citation:
Metz, Cade. “A.I. Shows Promise Assisting Physicians.” The New York Times, The New York Times, 11 Feb. 2019, www.nytimes.com/2019/02/11/health/artificial-intelligence-medical-diagnosis.html.
1 comment:
After reading this blog, I do agree that using AI to diagnose patients could be very helpful. Misdiagnosis is not something you hear about as a serious underlining issue often. According to Medical News Today, around twelve million people in the United States are misdiagnosed in outpatient clinics each year. This is a serious number, therefore adding the use of AI to help diagnose patients would only lead to positive outcomes. I do not think that we should ever rely on the use of AI alone because doctors, although they are prone to human error, have been through years of training and schooling, along with experience that a machine will never have. I think the better idea would be for the doctor and the AI system to work alongside each other. Although the AI system, would use all of the patients records to make a more accurate diagnoses, the issues with the confidentiality could be serious. If people do not feel comfortable sharing their health history, that is when a doctor with prior knowledge and years of experience would be more accurate.
One other topic to take into consideration is money. Out of those twelve million people we do not know what patients are able to afford top treatment and if their misdiagnosis is coming from cheaper practices or under what specific circumstances. If this is the case than the people who cannot afford the best treatment, will not be able to afford a practice with an expensive AI system. I’m sure in the first few years of the AI systems use it will be more expensive.
Overall, this idea has amazing potential and if the issues with the confidentiality could be solved and if they could come up with a reasonable price for patients, I think it will save a lot of lives. Using both the AI system and the doctor would make patients feel a lot more confident with what they are being diagnosed with
Post a Comment