As a Computer Scientist focusing on Artificial Intelligence, I find this video very arrogant towards medical professionals, exceptionally disrespectful and very misleading. A telescope is a tool used by astrologers, it does not however define astrology. In the same way that a stethoscope is a tool to help doctors reach a diagnosis, it is not the diagnosis itself. For those who do not understand Artificial Intelligence, it is the art of converting data representations into mathematical representations and then using statistical analysis to reach a probability as an outcome. this is subject to variable data which varies from patient to patient, it is subject to trained data which can be unbalanced and carry poor datasets. AI is often NARROW ai, it is not singularity (Terminator, Hal 9000), therefore to state who should you trust, a computer that has all your medical records or a doctor, the AI is not broad AI and is not going to diagnose a patient based on multiple medical specialities, at best it is going to make a probability (aka a guess) what the patient may have based on a series of inputs and cross reference that with the average from other known cases it has been trained with. Whilst one may argue that like medical students learning about medicine, I would state "similar, not the same" I am basing my current paper on Artificial Intelligence and classifying skin lesions based on features on a photograph, however that alone cannot predict skin cancer, given that a photo does not carry any patient history and often depends on datasets to train the model before using test data, more often than not datasets lack patient information, largely due to privacy laws and access to medical data, for instance, Race, Gender, Age, Familial History, Location (ie living within the tropics where UV exposure is much greater than within northern hemispheres), resolution of the photos taken, lighting, lifestyle (ie: use of artificial sun beds, smoking, cancer history, alcohol), even evolution of the lesion, ie timeline of changes as more often than not there is only 1 set of photos taken the same date and time of the patient, and as such whilst these models may outperform professionals in controlled test environments it is entirely different to then say they are better than doctors in real-world settings where data is not readily available) furthermore, there are risks that AI models can be "Overtrained" which then produces a risk of false positives, and like any medical test (even blood tests can produce false positives, if a patient is checked for things regularly it is possible for false positives to crop up on their medical records) If we look at other aspects of AI, you have the instance of Boeing 777Max autopilot correcting the pitch of the aircraft causing a stall and dismissing the pilot trying to correct the pitch, resulting in 2 major crashes that killed over 300 people, as such AI is never intended to replace human beings but rather assist, and the rule of thumb for any AI developer is to consider the moral implications, the legal implications, even with myself looking into skin cancer, there is a real possibility the AI may create false negative and false positives, and send one patient to a dermatologist for a biopsy, and turns out to be a benign lesion, whilst another result could have a result of a low probability of cancer and differ them from seeking medical help, and then risk metastasis as a result of late stage cancer development, this can prove fatal, especially with Melanoma, in which case a patient who would have sought help did not. Ai has amazing potential to be helpful in medicine and the practice of AI in medicine is one that dates back to the 1970s, but it is certainly NOT a replacement for a doctor and not one people should blindly place faith in autonomous decision.
AI & Machine learning is the future , it's good to se them contribute in our healthcare systems making it more flexible. It would be a boon in those areas where doctors can't reach easily ,like in India. It could change a lot in Radiology & surgery. Great Video. Thanks you so much.
While this video highlights the advantages of AI, to some extent it is disrespectful to doctors and healthcare workers in general. Saying that a computer can handle everything about your healthcare is an inaccurate statement, to say the least. In other aspects, this video is quite informational, but at many points, it disrespects the profession as a whole.
This is extremely simplistic and inaccurate way of how doctors come to a diagnosis. If your country has bad doctors or you have had a bad experience with doctors this is not an indicative of the quality of doctors overall or the quality of education we get. The medical data on which AI is trained is first and foremost in breach of HIPPA and the data could very well suffer security breaches and therefore cause many problems for many people. AI doesn't have any empathy, compassion, or human experience or ethics to supplement their written knowledge. Third, the data is in which it is trained is vastly contributed to it by English written medical records to which the vast majority is on white patients, we already have problems with limited data causing problems for non-white people and therefore this is just gonna take it to the next level. Overall, very simplistic, inaccurate, and almost manipulative of the general publics image and view of doctors and healthcare.
As a Computer Scientist focusing on Artificial Intelligence, I find this video very arrogant towards medical professionals, exceptionally disrespectful and very misleading.
A telescope is a tool used by astrologers, it does not however define astrology. In the same way that a stethoscope is a tool to help doctors reach a diagnosis, it is not the diagnosis itself.
For those who do not understand Artificial Intelligence, it is the art of converting data representations into mathematical representations and then using statistical analysis to reach a probability as an outcome. this is subject to variable data which varies from patient to patient, it is subject to trained data which can be unbalanced and carry poor datasets.
AI is often NARROW ai, it is not singularity (Terminator, Hal 9000), therefore to state who should you trust, a computer that has all your medical records or a doctor, the AI is not broad AI and is not going to diagnose a patient based on multiple medical specialities, at best it is going to make a probability (aka a guess) what the patient may have based on a series of inputs and cross reference that with the average from other known cases it has been trained with. Whilst one may argue that like medical students learning about medicine, I would state "similar, not the same"
I am basing my current paper on Artificial Intelligence and classifying skin lesions based on features on a photograph, however that alone cannot predict skin cancer, given that a photo does not carry any patient history and often depends on datasets to train the model before using test data, more often than not datasets lack patient information, largely due to privacy laws and access to medical data, for instance, Race, Gender, Age, Familial History, Location (ie living within the tropics where UV exposure is much greater than within northern hemispheres), resolution of the photos taken, lighting, lifestyle (ie: use of artificial sun beds, smoking, cancer history, alcohol), even evolution of the lesion, ie timeline of changes as more often than not there is only 1 set of photos taken the same date and time of the patient, and as such whilst these models may outperform professionals in controlled test environments it is entirely different to then say they are better than doctors in real-world settings where data is not readily available) furthermore, there are risks that AI models can be "Overtrained" which then produces a risk of false positives, and like any medical test (even blood tests can produce false positives, if a patient is checked for things regularly it is possible for false positives to crop up on their medical records)
If we look at other aspects of AI, you have the instance of Boeing 777Max autopilot correcting the pitch of the aircraft causing a stall and dismissing the pilot trying to correct the pitch, resulting in 2 major crashes that killed over 300 people, as such AI is never intended to replace human beings but rather assist, and the rule of thumb for any AI developer is to consider the moral implications, the legal implications, even with myself looking into skin cancer, there is a real possibility the AI may create false negative and false positives, and send one patient to a dermatologist for a biopsy, and turns out to be a benign lesion, whilst another result could have a result of a low probability of cancer and differ them from seeking medical help, and then risk metastasis as a result of late stage cancer development, this can prove fatal, especially with Melanoma, in which case a patient who would have sought help did not.
Ai has amazing potential to be helpful in medicine and the practice of AI in medicine is one that dates back to the 1970s, but it is certainly NOT a replacement for a doctor and not one people should blindly place faith in autonomous decision.
Very simplistic view about how Dr come into the diagnosis.
AI & Machine learning is the future , it's good to se them contribute in our healthcare systems making it more flexible. It would be a boon in those areas where doctors can't reach easily ,like in India. It could change a lot in Radiology & surgery. Great Video. Thanks you so much.
Good news.. now we have AI to reduce physician burnout as well.
While this video highlights the advantages of AI, to some extent it is disrespectful to doctors and healthcare workers in general. Saying that a computer can handle everything about your healthcare is an inaccurate statement, to say the least. In other aspects, this video is quite informational, but at many points, it disrespects the profession as a whole.
Video posted one year ago still I am the first comment 😂😂 but 3.6k views
Good sum up
This is extremely simplistic and inaccurate way of how doctors come to a diagnosis. If your country has bad doctors or you have had a bad experience with doctors this is not an indicative of the quality of doctors overall or the quality of education we get. The medical data on which AI is trained is first and foremost in breach of HIPPA and the data could very well suffer security breaches and therefore cause many problems for many people. AI doesn't have any empathy, compassion, or human experience or ethics to supplement their written knowledge. Third, the data is in which it is trained is vastly contributed to it by English written medical records to which the vast majority is on white patients, we already have problems with limited data causing problems for non-white people and therefore this is just gonna take it to the next level. Overall, very simplistic, inaccurate, and almost manipulative of the general publics image and view of doctors and healthcare.