From big data-driven diagnostics to personalized therapy: AI is expected to revolutionize medicine. What’s in for us patients? And in which fields do human doctors outdo artificial Intelligence?
Category
🗞
NewsTranscript
00:00 Imagine you're sick. Who would you trust when it comes to your diagnostics and therapy?
00:04 A human doctor or artificial intelligence?
00:07 AI is expected to revolutionize the field of medicine.
00:11 It can already sift through huge amounts of x-rays and analyze massive data sets
00:16 to find the best therapy for an individual patient.
00:19 So, will AI someday replace doctors?
00:22 AI systems can help with a more accurate diagnosis.
00:29 That's due to AI's ability to analyze huge amounts of data in the blink of an eye.
00:34 Just one example.
00:35 An AI can quickly compare MRI scans of any patient with others in a huge database
00:41 and spot early signs of cancer or Alzheimer's.
00:44 This quality makes AI a powerful tool for detecting and treating rare diseases as well.
00:49 Plus, AI can generate individualized treatment plans for you
00:53 and help you recover faster by analyzing all of your health data
00:57 and comparing it to other cases.
01:00 But also human doctors have their benefits.
01:02 An individual doctor draws on their training and experience
01:05 and hopefully looks at the patient with a holistic approach.
01:09 That means he or she would take many factors into account.
01:12 AI can only "see" what it's been trained to see.
01:15 It's very precise and quick on that aspect, but might overlook others.
01:20 But it's not about deciding between AI or doctor, of course.
01:23 The question is rather, when does every doctor get to benefit from AI's assistance?
01:28 So, what do we need to get started?
01:30 Medical AI systems need a lot of data.
01:37 And the more, the better.
01:39 Because artificial intelligence is trained to recognize patterns by using processed data sets.
01:45 Your biometric data is also needed.
01:47 By assembling information from your medical record, your smartphone or health gadgets,
01:52 AI can create a digital twin of you.
01:55 Such a virtual doppelganger can not only get you a perfectly tailored individual treatment,
02:00 it could also be used for developing new drugs.
02:03 Drugs could be tested on digital twins before starting expensive human trials.
02:07 It would be cheaper, quicker and safer.
02:10 Today, it takes an average of 10 years and billions of euros to develop a new drug.
02:16 Using AI to predict how potential drugs behave in the body
02:19 could really cut down on the time and money needed.
02:22 Dead-end compounds could be identified and discarded right away.
02:26 There's a wide range for possible applications for AI in the health sector.
02:30 But when will it be widely used?
02:32 Well, there are some obstacles.
02:37 Data security is a big issue.
02:39 Your medical record contains very sensitive private information
02:43 and could be misused if it falls into the wrong hands.
02:46 Also, keep in mind that data fed into an AI system can't just be erased.
02:51 An AI learns from the input you give it.
02:54 It's not a simple database that lets you delete specific entries.
02:57 Just like you can't erase a memory from your brain, data fed to an AI system stays in there.
03:03 And then there's ethical implications.
03:05 Medical datasets need to be put together in a truly representative way.
03:09 AI only learns with the data it is fed.
03:12 For example, in case women were underrepresented in the training data,
03:16 a diagnosis could be less accurate for them than for men.
03:19 It could even be wrong.
03:20 This gender gap has a long history in medicine.
03:23 Drug trials were often conducted only on men,
03:26 even though the same drugs were prescribed to women.
03:28 This led to ineffective treatment for them.
03:31 Drawing on that biased data to train AIs runs the danger of replicating the issue.
03:36 Plus, who is liable if the AI suggests a wrong diagnosis or treatment?
03:41 Is it the doctor who used it or the company who deployed the AI system?
03:45 Legal questions like that still need to be answered.
03:48 I think it needs to be us humans taking control and responsibility.
03:52 But AI can be of great help.
03:54 What's your view on AI in medicine?
03:56 Are you optimistic about its potential?
03:58 Or would you rather rely on your human doctor?
04:01 That's all from me today.
04:02 Which topics would you like us to cover next?
04:05 Let us know.
04:06 Bye and see you soon.