AI in the medical space

With the outbreak of the coronavirus in China, the world’s attention has been drawn towards the future of health care, specifically disease prevention.

Since its outbreak in December the virus has been unstoppable. It has reached pandemic proportions and the World Health Organisation has declared it a global public health emergency.

A major concern regarding coronavirus is that it could easily spread to countries with poorly developed health-care systems and South Africa is on that list of countries.

While scientists scramble to find a cure, a big part of international efforts is to contain the virus through early detection and quarantining infected patients. Prevention is better than cure.

What makes the situation worse, is that Chinese scientists have positively diagnosed a newborn baby with the virus, just 30 hours after birth.

This raises fear among scientists that the illness could have been propagated from mother to child before birth, making it even harder to combat.

What is particularly frightening about coronavirus is that despite all our advancements in medical technology, we are unable to stop its spread, nearly two months after its outbreak.

Coronavirus has confirmed what we have known for a long time: that we need a new breakthrough in the field of disease prevention, one that will take health care to a new level.

Fortunately, that breakthrough is on the horizon.

For some time now, scientists have been exploring the potential of artificial intelligence, or AI, in the medical field.

AI has made tremendous progress in a number of fields, and if that level of success could be replicated in the medical field, we could have that major breakthrough we so desperately need.

In one such case, researchers from Google Health and the Imperial College in London developed and trained an AI system to detect breast cancer by studying mammograms, which are X-rays of the breasts.

The system was shown mammograms from tens of thousands of women. Not surprisingly, it learned how to detect breast cancer with a high level of accuracy.

What was surprising, though, was that the AI outperformed human radiologists by a wide margin.

In a normal breast cancer detection process, two radiologists team up to study each mammogram for signs of cancer. This is to ensure that there are two opinions up front. If the two concur on a finding, it is taken as the final diagnostic.

If the two radiologists are unable to come to an agreement, then a third radiologist is brought in to settle the matter.

All told, the process is time-consuming, and not that accurate. Errors and misdiagnosis are common, with some studies showing that 20% of breast cancer cases go undetected. That is an extremely high number.

What is more remarkable is that the system had far less information to work with than the human radiologists.

The AI system was only given the mammograms to look at, and nothing else – which makes its achievement even greater.

Researchers at Google believe that as they find ways to input the additional data into the AI system, it will become much more accurate still. Apart from reading the mammograms, the AI system will deliver another huge benefit – it will store, categorise and analyse the data it receives, with its diagnostics.

As it reads more and more information from women all over the world, it will use big data analysis to draw wider conclusions about global breast cancer patterns.

Using this data, it might unearth previously undetected patterns, such as the prevalence of breast cancer in certain countries. This will enable doctors to prevent further cases by advising people in high-risk areas to avoid certain foods or behaviours, for example.

This type of big data analysis has already been used very effectively in the US to detect early outbreaks of sexually transmitted diseases.

Although we are in the early stages of using AI to improve health care, it is already showing tremendous promise.

While medical care will always be a human activity, AI will prove to be an invaluable ally to medical professionals.

Share This Post
Have your say!

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

This site uses Akismet to reduce spam. Learn how your comment data is processed.