Can AI be used in healthcare

predictions of the AI healthcare market

Disease diagnosis

40–80 thousand people die every year just because of misdiagnosis, how can we ignore that number? If we want to use AI for self-driving cars , then surely a good use for it would be to save lives. AI can be used in diagnosis, with many benefits, such as: we don’t have to spend time and resources training medical professionals, and we reduce human error in diagnosis. We can also get diagnosis’ earlier than we would with a human doctor, early diagnosis is projected to have saved 10 thousand lives! In 2017, there were 1.566 physicians for 1000 people globally, with Germany being the highest with 4.3 physicians per 1000 people. This means that there is an insane amount of pressure on the healthcare system. By using image processing, we can assist in the diagnosis of diseases which require medical imaging. This is done by supplying the algorithm with thousands of cases (confirmed and suspected) to train the algorithm so that it can then predict these cases.

How AI disease diagnosis happens right now

Some examples of where AI is used in medical imaging are as follows. Firstly, AI in medical imaging is very useful for revealing cardiovascular abnormalities, AI will help speed up the process of image analysis when screening for cardiovascular abnormalities due to it being a repetitive task which can easily be automated. AI can also help to predict alzheimer’s disease years prior to it happening, it does this by noticing possible metabolic brain changes. By 2030, AI will access multiple sources of data to reveal patterns in disease and aid treatment and care. Healthcare systems will be able to predict an individual’s risk of certain diseases and suggest preventive measures. AI will help reduce waiting times for patients and improve efficiency in hospitals and health systems. AI is also phenomenal in cancer detection, and is even doing better than some medical professionals in detecting breast cancer, we can also use AI to check and see if treatments are going well. The final use that I am going to state is that AI can help plan surgeries to reduce the time it would take for surgeries to take place. Some of the leading companies in this are: IBM Watson, one of the pioneers in this field of AI. “IBM aims for fast processing of medical images and to interpret the data efficiently with information from various databases.” Another company in this field is Butterfly network, which aims to bring a different perspective on medical imaging with both hardware and software solutions. There are also many more that are doing amazing work.

Medicine and prescriptions

Companies working in the industry of AI drug discovery and development

To deviate from diagnostics, I want to focus on another part of AI which is AI prescriptions. In the United States, between 7,000 and 9,000 patients die from medication errors every year. That can be avoided by using AI to determine the kind of prescription that should be given to an individual based on what is most likely to be effective. GPT-3 is quite competent at prescribing medicine. GPT-3 is a language processing model with the largest database of training, it can answer medical questions, diagnose asthma and prescribe medicines. While GTP-3 is amazing, there are also other amazing things out there, such as Google’s prescribing model which can predict what a physician would prescribe with 75% accuracy. Training AI to prescribe medicine can be very effective, especially for reducing human error and saving lives, but should we stop here, I would argue that AI has the potential to do a lot better in medicine, and many researchers agree with me.

Diagram of how AI works in healthcare

When I talk about AI not reaching its limit at normal prescriptions, I am talking about precision medicine. Precision medicine is medicine for a specific individual that takes into account an individual’s genes, lifestyle and environment. This helps doctors to create a personalized treatment that should be more effective than a “blanketed approach” or a “one size fits all approach”. Precision medicine has been here since before it was involved in AI, for example in a blood transfusion a person is not just given any blood, but they are matched with the blood of a donor to reduce risk of any complications. Precision medicine wants to provide the most tailored medicine for any range of health issues, and can also be used to keep healthy patients even healthier. It tries to know as much about a patient as possible, by understanding things like the genome, metabolism and sometimes even look at someone’s social media. Some of the leading precision medicine organizations in the UK are: the Babraham Institute, which is a leader in epigenetics, Cancer research UK and CPRD which is funded by the NHS.

The challenges that AI faces in drug discovery

Cell simulations

Even though everything that I have been talking about looks like it will be used after 20 years time, it is undeniable that artificial intelligence is already everywhere and not just a science fiction concept, as soon as you open your phone to scroll on things like TikTok and Youtube, there is an algorithm deciding what videos you should watch. The future may be filled with self-driving cars and the things that I have been talking about so far, but personally, the thing that I find mind-blowing are integrated cell models. The goal for these cell models is to see how cells are structured in 3D and how they develop over time. To do this, you would have to quantify shapes from major structures in cells that you get using fluorescent microscopy, fluorescent microscopy allows you to acquire high quality 3D images but does not allow for you to analyze many cell structures at once due to the damage caused by an excessive amount of laser light. Normally, when creating a model like this, you would look at the cell membrane, the DNA and one endogenously tagged protein designed to reveal the location and morphology of a single cellular structure. You could also create a 3D probabilistic model of a cell which predicts the likely structure of a cell(wish I could predict the likely outcome of my math scores, but cells are cool too!)

How we create integrated cell models currently

We can use these 3D cell models to understand how cells develop and react to specific things. They are also used to understand disease mechanisms and discover drug therapeutics. The process may involve deriving 3D cultures, such as cancer organoids, from patients. The 3D cultures can be used to screen for small molecule drugs or genetically manipulated to understand disease pathways. We can use these models to test new medicines and make them more effective. The ​Allen Institute For Cell Science has created a 3D cell model that you can explore on their website and has also created a 3D probabilistic cell model that we can explore as well. As well as this they have some of their code available to explore online.

Some integrated cell models

Looking at all of the different uses of AI in just the healthcare industry, AI is probably going to have a massive impact on the world. AI will most probably save a lot of lives in things like diagnostics, prescriptions, precision medicine and cell models! The future of healthcare looks so bright, hopefully you enjoyed reading that article and I didn’t make it a total snooze-fest!

A summery of what aplpications AI has in healthcare

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store