Share this post on:

We find artificial intelligence (AI) in smartphones, cars and homes. There are those who fear it for its varied and uncontrollable uses, but there are those who exalt it considering only the advantages.

However, we are aware the fact that future generations are destined to coexist with machines, which will perhaps decide for us and will automate some of the major human routines. Not only everyday life, but also medicine, a good example is the recent use of AI in cancer forecasting, in the diagnosis and treatment of tumours.

More and more often in scientific conferences, we hear about artificial intelligence against tumours, in particular how machines can help diagnostics. AI tries to reproduce human mental processes using algorithms and computers, while machine learning is the branch of AI that deals with computers and software to process big data. Both serve to fight tumours, but how?

Let’s find out why investing in e-health is an important choice for medicine. 


Artificial intelligence aims to create software or hardware capable of solving problems as it were a thinking human being, ranging from the ability of distinguishing people’s faces to that autonomously processing a huge amount of data. It is believed that around 2050 artificial intelligence will reach human intelligence, revolutionising oncology, especially in the diagnosis and treatment of the disease.

Google researchers have inserted a large number of images acquired by CAT on a computer, with the aim of getting it used to recognising the lung lesions typical of a certain kind of tumour. The result was surprising: the computer developed an algorithm able to assess the presence or absence of this disease in a patient, obtaining a higher percentage of correct diagnoses than an expert radiologist. This shows that artificial intelligence against tumours can act more effectively than humans, especially in reading images related to this type of disease.

Obviously, it does not diminish the work of professional radiologists, since the innovative medical device and the radiologist are equivalent when it is necessary to compare a recent CT scan with an older one and find the differences. However, this discovery can optimise the screening process for the tumour, using the computer as an assistant and not as a substitute.


With regard to the critical issues and problems relating to the use of AI in medicine, the first difficulty concerns the reliability of the algorithms in the medical field for all patient groups. It is necessary to verify that the data produced by the various institutions are heterogeneous and of quality.

Above all, it is necessary to understand what is the judgment logic of machine learning, because doctors make clinical decisions based on objective parameters, while the parameters that influence the creation of the algorithm may not be evident. As for the protection of patient privacy, however, institutions still have to deal with regulation.

Although they enable to increase productivity and improve results, artificial intelligence and machine learning are not infallible, therefore it must be considered that they have some limits too and must be treated responsibly by health systems. Let’s see together the other technological developments useful to understand how to improve artificial intelligence in oncology:


Artificial intelligence also helps medicine because it increasingly supports doctors in making decisions. In this case, we talk about disrupt healthcare.

In Italy about a thousand cancers are diagnosed a year, but the cure rates improve (63% of women and 57% of men, according to data from the Italian Cancer Registers Association). Certainly, it is largely due to screening campaigns, increased awareness and targeted treatments to reduce mortality.

The DeepMind project uses medical data to provide better health services, such as the ability to process a lot of medical information in a short time thanks to machine learning. DeepMind, together with the British National Health Service, works to speed up the times related to radiotherapy thanks to a “segmentation” process, with the aim of decreasing these times by up one hour for some types of cancer.

To do this, it is necessary to perform a real artificial intelligence training in the field of treatments, using a large amount of scanned clinical images useful for the software to identify those areas of tumour tissue and distinguish them from healthy ones. Once the problem is recognised, the AI recommends the right actions to be undertaken.

IBM’s Watson project, used in multiple contexts, wants to be the turning point in the fight against cancer thanks to artificial intelligence and with an accuracy of 96%. He could do this by assisting oncologists to suggest more precise diagnoses and treatments. So far 55 hospitals worldwide have adopted Watson, especially with regard to breast cancer, ovarian cancer and gastric cancer.

Artificial intelligence and human beings in comparison

Neurology Genetics recently compared IMB’s Watson “thinking” with that of a team of medical specialists. The aim of the research was to diagnose and treat a brain tumour in a 76-year-old patient. In a very short time, artificial intelligence recognised the type of tumour and made a proposal for treatment, while the team of doctors would have taken much longer, but they probably would have proposed a more effective treatment. Therefore, even if artificial intelligence is undoubtedly faster in diagnosis and therapy, doctors’ decisions have managed to make connections that the machine is not yet able to do.

In Italy, the Human Technopole research centre, still in development, has a Data Science centre that will develop patient data analysis algorithms.

Will artificial intelligence therefore replace doctors in the fight against cancer? Obviously not, in fact IBM’s Watson is still not able to distinguish different forms of cancer. The software recommends treatments based on its knowledge, which is based in turn on human knowledge. Watson for Oncology is therefore still an early stage of development, but soon it may be able to guide doctors in the treatment of 12 different types of cancer.

Although it does not create new knowledge on its own, Watson can learn and has enormous potential yet to be fully explored. In a certain sense, it’s like AI is affected by a cognitive disease and even if it assimilates some skills, it is not able to apply them in other sectors.

In Australia, researchers are using mammogram data to “teach” AI to identify patterns that a human doctor could elude. The results are accurate, since the technology is the same used by Siri, the Apple assistant and other hi-tech devices.

Looking for e-health project partners? You found it in IPPOCRATE AS!
This site uses cookies to improve users' browsing experience and to collect information on the use of the site.