Artificial intelligence in healthcare

What is artificial intelligence?

Medical research is going through a digital revolution associated with the abundance of information and our ability to collect, store and, above all, process it. These possibilities are linked to the capabilities of computers, which today are almost limitless: they can analyze billions of data at sufficient speed and compare many parameters, where the human brain cannot handle more than 6 or 7 variables at a time. The development of artificial intelligence (AI) associated with the skill of doctors is therefore very promising in terms of health.

But what is AI? This field depends on two important elements: algorithms and big data.

Algorithm

An algorithm is a series of operations and / or formal rules performed by a computer program to solve a given type of problem. More concretely, this computer program Imitate the mode of reasoning that calls for intelligence.

There are AI programs where tasks will be predetermined by humans. For example, it is a program that knows how to play chess because it provides the computing power needed to predict the rules of the game and its preferred outcomes.

There is also an AI program Learning, Where the system learns on its own, until it is given the right practice and above all the right answer. For example, if the program is provided with photos of many objects and their associated identification, it will be able to distinguish a car from an aircraft in any subsequent new photo. Some programs, if given enough data, are able to discover their own similarities even without any prior guidance. We then speak of deep learning or deep learning.

Big data, or big data

Another essential element for AI: big data or big data. In the field of health, digital data is on the rise: health insurance files, cause of death or national health data system files, patients’ medical files (test and imaging results, surgical procedure reports, therapeutic protocols, etc.). Which adds data from research programs.

So there are many issues surrounding all this information to allow AI to make the best use of it. This includes ensuring their accuracy, their real-life representation, but also the quality and sharing of their organization (which requires a fair balance between data privacy and access to this data).

Examples of the application of artificial intelligence in the field of health

In oncology

Patient data structure

The ConSoRe project (Continuum Care – for research), founded by Unicanser, a network of cancer-fighting centers, aims to organize the huge data collected in the field of oncology. It is a system whose goal is to collect, analyze and structure this data. Associated with a search engine, it allows doctors to identify patients who meet certain search criteria, visualize their disease and the evolution of their treatment, identify rare cancer files, or even determine if similar cases have already been treated elsewhere. This data can then be processed by various algorithms.

Give the benefit of diagnosis

It is probably in the case of diagnostics that AI will first disrupt oncology. Several research teams around the world have already proven that programs are able to detect melanoma, a serious form of skin cancer, or breast cancer more quickly than a doctor can, even when there is an atypical tumor.

Another example: In the framework of Plan France Médecine Génomique 2025, research projects in bioinformatics (data applied in biology) attempt to find a correlation between genome analysis and clinical manifestations of cancer. In this way, they hope to be able to genetically diagnose certain cancers, making it possible to personalize treatment and open up new therapeutic avenues.

Guide therapeutic management

A team at the Cordeliers Research Center in Paris has developed an algorithm that predetermines that patients with any rectal cancer will respond fully to radiotherapy, thus avoiding unnecessary surgery.

As part of the European Disease Project (Decision Support and Information Management System for Breast Cancer), which involved AP-HP and LIMICS Laboratories, a therapeutic decision support program for women with breast cancer has been developed, based on best practice recommendations. Also on the experience of all the decisions taken during the multilateral consultation meeting.

Assistance in the prediction of cardiovascular field and replacement

Another area where AI is becoming dynamic is cardiovascular health. For example, a team at University College London is developing an algorithm for predicting the risk of death after a myocardial infarction or stroke based on analysis of the heart’s magnetic resonance imaging data. In France, Samu is working on a French adaptation of a Danish AI program that is able to help 15 operators detect cardiac arrest, thanks to real-time, verbal signal (keyword) and non-verbal (voice) analysis. , Rhythm of breathing, etc.).

In the case of transplantation, an international study conducted by a French team validated a universal algorithm for predicting the risk of a transplanted kidney failure in a transplanted patient. One step further to improve patient follow-up and optimize the development of new immunosuppressive therapies.

Support for gynecological-obstetric diagnosis

Medical imaging data analysis is one of the most searched fields by AI. The European SUOG (Smart Ultrasound in Obstetrics and Gynecology) project, led by teams from Sorbonne-University, Insarum and AP-HP in France, aims to use AI to improve pregnancy monitoring. In case of suspected pathology the pictures that he has to take for diagnosis are able to advise the practitioner in real time.

Prevent mental health

There are also mental health research activities, especially in the area of ​​prevention. The Psychiatry Project (Early Intervention of Psychosis: Towards Preventive and Personalized Psychiatry), led by Marie-Odil Krebs, a physician and researcher at the University of Paris, specifically develops an AI that allows early detection of schizophrenia or chronic psychosis. Preventive and personalized mental care up. To identify individuals at risk, AI will intervene by specifically modeling knowledge about schizophrenia and identifying a set of biomarkers used through a learning algorithm.

Artificial Intelligence: A tool that remains to be mastered

As you can see, the purpose of researchers working on AI projects is not to replace doctors, but to help them make better diagnoses, better treatments, and better risk predictions.

Jean Charlotte, a researcher at the Laboratory of Medical Informatics and Knowledge Engineering for e-Health in Paris, explains, “There is a lot of evidence today that AI can be beneficial to health. But like any new tool, it will change the way things are organized. However, the more sophisticated a tool is, the more in-depth it will change the previous organization. A

There is no doubt, therefore, that AI will gradually become established in the field of health, with the consequence of patient / doctor relationships and the emergence of new ethical issues.

Source

Leave a Comment