ARTIFICIAL INTELLIGENCE IN 2020 PANDEMIC
‘With
great power comes great responsibility.’ Artificial Intelligence, one of the
most recent and powerful inventions of humankind, has come into play to fight
against the infectious lethal COVID-19 virus.
On
February 29, 2020, Washington state confirmed its first death of a patient
suffering from the coronavirus. Just 2 weeks later, the state reported a
shocking statistic: 40 deaths and more than 600 confirmed cases of the virus.
Washingtonians started to call their local hospitals for answering questions
about the symptoms and even showing up at ER for medical check-ups. This influx
of calls and ‘patients’ not only overloaded local hospitals’ services, but also
put lives of many people under risk since they could not reach the health
experts to receive appropriate medical advice/treatments. To address this
problem, one Washington hospital -- Providence St. Joseph Health-- decided to
build an AI chatbot, under Microsoft services to help answering the
anxious worried people and directing the infected/high-risk to the appropriate
resources. This helps hindering patients from overloading the hotline as well
as the ER service. Within
the first few days, the chatbot ‘has made more than 150000 messages and served more than 40000
sessions.’ With the success of the
chatbot, many hospitals in the U.S are using/building the AI Chatbot to help
screening the ones who might be infected and reassuring those who are not. A
few representative hospitals including Boston’s Partners HealthCare, Tampa’s
General Hospital, Cleveland Clinic, or Illinois’ OSF Healthcare System, and
many more as around 300000 people have already been confirmed as patients
of this virus.
Meanwhile in Europe, which is also severely impacted by the virus, healthcare centers start to use the power of Artificial Intelligence in CT scanning. Due to the lack of testing kits and the recent questions on testing kits from China, which has been the main suppliers of masks/gloves/protection suits/testing kits for many European countries, detecting infected people through CT scans of their lungs, as an alternative to using testing kits, recently drew experts’ attention. However, just like the previous logistical problem in the U.S, the influx of scanned lung images from an exponentially increasing number of patients may create a big burden on clinicians and radiologists, of whom number can only be trained and available in a linear scale. One of the solutions is to speed up the time it takes to read scanned images yet remaining, or increasing, the accuracy of the final diagnostic. Enters AI. “ While a manual read of a CT scan can take up to 15 minutes, AI can finish reading the image in 10 seconds, ” said Erik Ranschaert, president of the European Society of Medical Imaging Informatics (EuSoMII).
This new AI-related alternative seems to be the new light in the
effort of accurately detecting the infected with a big and increasing AI
community. One AI company, RADLogics, recently claimed that they have
successfully built a model that had the detection accuracy of up to 98% in
their pre-print research.
It is also reported
that around 30 health organizations across Europe are collaborating to provide
data sets and to train the deep-learning model through the ‘AI processing
radiological images’ server Quibim and the Robovision AI (RVAI) software.
With these innovative movements, the healthcare system may come up with a more
efficient way to cover a large amount of the population who are infected.
(there are debates, since taking CT scan for one person requires an enclosed
room with staff and the patient, and it’s time-consuming to sanitize all the
equipment before advancing to take CT scan of the next person).
During
this 2020 pandemic, Artificial Intelligence has proved to be a powerful and
beneficial technology to help us address problems involving logistics and
processing large amount of data for predictions. Despite current debates about
how AI could threaten our existence, this current application of AI in fighting
against the COVID-19 Virus has shown that 1) AI is a powerful and controlled
tool, and 2) whether the tool is helpful or harmful is up to the ones who are
in using it.
I'm curious what people's reactions to the AI are. Some people might feel irritated or less trusting since they are dealing with a machine instead of a person, and I could image that might impact the development of AI in the medical field.
ReplyDeleteI think this is an interesting point. I mean, I don't like chatting with a bot too. Furthermore, there are also health experts who do not fully trust in this new technology. However, in this situation, I think people have to except it because there is no other way around: hospitals don't have enough facility and human resources. So, if AI could success in helping hospitals and patients during this time, it will be trusted and used further in medical field.
DeleteI think that this is a great way of providing tests for people. As of right now, the United States struggles to meet the demand of testing because of administrative power and how long it takes to complete the tests. By having AI help with the testing, this would greatly reduce the time for the results as you have stated in your post.
ReplyDeleteAbsolutely! Honestly I feel a bit gutted about the accuracy that the bot/AI offers, but I also believe that the developers and hospitals have done their best at the end to provide the tools so the statistics should be accurate.
DeleteI'm so fascinated by how fast these applications were developed and put into use. The AI chatbot must have helped tons of people who were freaking out about whether they carried the virus. Have you tried talking to the bots? They give very useful information but was not as friendly as I thought they would be.
ReplyDeleteIt was my bad not trying to talk to the bot before publishing the post! I don't know how they built the bot, but it could be similar to the Eliza bot that we studied earlier in the course (and we all know how unfriendly it could be!). I'll definitely give a try talking to it.
DeleteI think they should have better technology for building chatbots now? For the one you mentioned above, it actually can save a bunch of information and response more personalized to me than Eliza, who actually can just repeat what we said in the immediate question.
DeleteI did some research and I think they haven't published any paper/information on this yet...
DeleteIn my opinion, no matter how powerful the AI is, we, as humans, have protect ourselves and our families by social distancing. If the AI made any false negative diagnosis, that would cause a big trouble to our community. Therefore, I agree that AI is a powerful and controlled TOOL. We have to be responsible for our own existence.
ReplyDelete