Artificial intelligence continues to take huge strides in the medical sector where over the next few years it will become increasingly important in treating many illnesses, such as tumours and metabolic disorders. A future in which healthcare employees and new computer programmes can operate side by side, combining their expertise to provide earlier and improved diagnoses and a greater degree of accuracy in the monitoring of treatments. But what are the specific benefits of AI when it comes to looking after our health? And when we talk about artificial intelligence in medicine what exactly do we mean?

Algorithms in white coats

Like in all professions, doctors acquire their expertise through experience. Examining ultrasounds, forming clinical pictures beginning with abnormal blood test results, charting the medical history of patients: the richer this “archive” of information, the more “expert” the doctor can claim to be.

But if rather than a single doctor we considered numerous doctors or, even better, all the doctors in the world or even, going back in time, every doctor that has ever existed, we would have an immense database of information that would be too large to use.

This is where digitalisation and information technology come in. In fact, nowadays we have archiving systems that enable us to store and manage quantities of data that we wouldn’t have thought possible just ten years ago. Another area undergoing rapid development is the field of machine learning, which gives computer systems the ability to learn without being explicitly programmed. In this way, they are able to study data and recognise recurring patterns therein, which they then use as raw materials to make their own predictions.

The clinical eye of the computer on the skin

Many of us will have had our moles mapped: a dermatologist surveys all the brown marks on our skin and examines those which, on the basis of a visual analysis (size, shape and colour), appear more suspicious.

Well a team of researchers from Stanford has just developed an automated version of this check-up, creating an algorithm for the diagnosis of skin cancer. To do this, the scientists created a database of nearly 130,000 images of confirmed cases of skin cancer and trained the system to recognise potentially dangerous moles in new images. The results? Very positive from the outset with the algorithm matching the performance of human dermatologists.

These results could speed up and also lower the cost of the early diagnosis of a lesion which – let’s remember – is one of the deadliest forms of skin cancer. And, taking it to the next level, the next step could be a smartphone app equipped with the algorithm that is capable of analysing our skin using the phone’s camera.

Beating breast cancer

Another interesting result, this time achieved by a research group from the University of Budapest, is the development of a system for the detection and evaluation of lesions in mammograms. As with the previous case study, the algorithm is able to identify suspicious regions in mammograms of its own accord.

With a level of sensitivity that enables it to detect around 90% of malignant lesions in the dataset and a very low margin of error in terms of false positives.

Not just cancer, not just diagnosis

As alluded to, the application of AI in medicine is not only targeted at the early diagnosis of tumours. A recent study on tuberculosis, a serious illness of the respiratory tract, has shown for example that a computer “trained” with hundreds of lung x-rays (of both healthy patients and patients with the illness) can improve the accuracy of its diagnoses as more data is added, reaching levels of reliability that are similar, if not superior, to those of human doctors.

As well as diagnosis, AI can also monitor pharmacological treatments, measuring the effects of a treatment (a course of chemotherapy, for example) on a specific ailment. For example, a Manchester University study subjected a number of tumours to very sophisticated measurements using the “eyes” of a computer, which assessed their evolution during the treatment.

In this case, the approach taken was quite unusual, mirroring a technology used by scientists to study the geography of Mars and assess its craters and dunes.