AI: a powerful tool at the service of health

In a digital world, algorithms underpin many of the systems we are used to living with. Now, artificial intelligence (AI) is everywhere in our daily lives. On the internet, when we query a search engine, when we ask it to translate a word or a sentence, when we surf on social networks. When writing an e-mail, when our mailbox suggests the text to us. With our smartphone, when it unlocks through facial recognition. But also in cars capable of anticipating obstacles, braking in an emergency or showing us the best route to follow, etc.

Unsurprisingly, medicine also benefits from these technologies. AI is even expanding there. A very complex discipline in itself, medicine is all the more so today as it produces a large volume of data. A single person, through their characteristics, illnesses, genetic code, laboratory analyses, imagery, etc., represents a large amount of medical information. As for the weight of scientific literature, it is also substantial, illustrates Prof. Christian Lovis, chief physician of the Department of Medical Information Sciences at the University Hospitals of Geneva (HUG): “Every day, more than 3000 publications are indexed in pubmedthe great computerized library of medicine.”

An augmented intelligence

As sophisticated as it may be, the human brain is not able to manage and analyze all this data on its own. “His memory capacity is reduced and he is subject to a certain fatigue which can impact his performance. This is why he needs to use tools to go beyond his limits,” explains Professor Antoine Geissbuhler, chief physician of the HUG Cyberhealth and Telemedicine Department. This is the whole purpose of artificial intelligence. Attempting to mimic human intelligence, AI is a set of technologies based on the creation and application of algorithms (precise strings of instructions that must be followed in order). Its purpose is to allow computers to perform tasks as human beings would. Ideally, “AI and human intelligence complement each other and it is then possible to speak of augmented intelligence”, describes Professor Geissbuhler.

How can AI help physicians in their practice? And above all, how can it improve the health and care of patients? In theory, the fields of application and uses of AI are endless. But today, in medicine, it is mainly used to aid diagnosis, predict risks and individualize treatments. And for good reason, the speed and precision of the machines represent real assets, confirms Jean-François Pradeau, Director of Information Systems: “Algorithms, because they can go very fast and analyze a lot of information, can be of great support in making a diagnosis or determining a risk.”

In our hands

Many of these tools are intended for professionals, but some of them are developed for anyone who has questions about their health. “The symptom checkers (or symptom checkers), for example, give the possibility of describing a symptom and obtaining a first opinion, as a kind of preliminary diagnosis to know whether to consult a pharmacist, his or her treating doctor or the emergencies”, illustrates Professor Geissbuhler. And to continue: “Some applications already make it possible to take a photo of a mole and to query software to find out if there is a risk of melanoma. The idea being that an appointment with the dermatologist can be obtained quickly in case of suspicion. With tools capable of categorizing the risk, the entire management strategy could be accelerated. In regions where medical density is low, these are very promising instruments.”

To the hospital

At HUG, for example, “AI has already entered into everyday life and is in full deployment”, declares Pre Laura Rubbia-Brandt, physician-head of the Diagnostic Department and the Clinical Pathology Service at HUG. Most doctors still use it occasionally through the digitized patient file or to find out about potential drug interactions, etc. In some departments, however, it is routinely used for more complex tasks. In radiology and clinical pathology, via digital pathology technology in particular, artificial intelligence programs are able to improve diagnostic performance by directing the doctor’s attention: “With great speed and control, the AI ​​selects the right images to analyze and allows very precise measurements of various parameters. We can thus accurately detect the presence of metastases in a lymph node or assess the level of proliferation of a tumour, for example”, indicates the professor. In oncology, where precision is very important to personalize treatments, AI could also offer added value.

In theory, any specialty, in addition to diagnostic medical disciplines, could in the future benefit from such software with ultra-powerful memory and computing capabilities. Always with the aim of improving the quality of care. At HUG, several research groups are working on AI programs to assess their relevance in specific applications. “We have, internally, the capacity and the skills to develop our own solutions, in collaboration with the major schools”, emphasizes Jean-François Pradeau.

Unload the human mind

Despite the power of these tools, they cannot replace the expertise of doctors. “We are in a logic of increase and not of replacement. A bit like a stethoscope that amplifies sound”, illustrates Professor Geissbuhler. However, certain tasks could be entrusted to the AI ​​because of its speed in processing and categorizing information, justifies Professor Lovis: “This concerns anything that is repetitive, which requires a lot of rigor and attention, but little thought. The automation of certain actions carried out in the laboratory, for example, would relieve the human mind.

Cumbersome and time-consuming, administrative management can also benefit from the intelligence of computers. Especially when it comes to fetching information from different places, synthesizing it, producing reports or billing for services received during a consultation or a hospital stay. Medical staff have everything to gain from this, underlines Prof. Geissbuhler: “Doctors can thus concentrate on what concerns human qualities such as listening, empathy and communication, which are at the heart of the therapeutic relationship. »

Thus, in addition to human intelligence, artificial intelligence offers real potential for improving care. Provided it is well designed, carefully evaluated, but also framed.

AI words

Algorithm: string of precise instructions that must be followed in order for a defined purpose, such as in a cooking recipe. Thanks to a programming language, the algorithm is translated into a program executable by a computer.

machine learning or machine learning: form of artificial intelligence based on mathematical and statistical approaches to enable computers to learn to perform certain predictive tasks.

deep learning or deep learning: type of artificial intelligence derived from machine learning. Here, the machine is able to learn by itself based on a neural network.

Neural network: type of algorithm inspired by the functioning of biological neurons. Based on previously recorded data, such as medical imagery, it is able to analyze new images and learn in order to make predictions.

_________

Similar Posts

Leave a Reply

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