Interview with Mustapha Ouladsine and Stéphane Delliaux.

Mustapha OULADSINE,

Professor of Universities exceptional class in Computer and Automatic Engineering, Vice-president of Aix-Marseille University delegated to digital and artificial intelligence for research, ex-Director of the Computer and Systems Laboratory from 2018-2020 and ex-Director of the Information Sciences and Systems Laboratory from 2008 to 2018. He is a specialist in data processing and AI for diagnostic assistance. Leader of the Laënnec Institute in Digital Sciences and Artificial Intelligence for Health at Aix-Marseille University. He is also co-founder and scientific advisor for the start-up WitMonki.

Doctor Stéphane Delliaux,

 is University Conference Professor – Hospital Practitioner at the Marseille University Hospital Center, Marseille, France. Specialized in Anesthesiology – Surgical resuscitation and human medical physiology, he works at the CardioVascular and Nutrition Research Center (C2VN) of the Faculty of Medical and Paramedical Sciences of Marseille as well as at the Laboratory of Functional Respiratory Explorations and Exercise of the North Hospital of Marseille. Co-leader of the Institut Laënnec in Digital Sciences and Artificial Intelligence for Health at Aix Marseille University, he is also co-founder and scientific and medical advisor for the start-up WitMonki.

1- Lte magazine (Khaouja) question:  What place does AI have in the health system today? and how will it evolve in the future with new technologies?

Answer from Mr. Ouladsine: In the age of digital and artificial intelligence, healthcare and medical practice have everything to benefit from technological advances in the digital world. AI in healthcare must leave our research laboratories for the benefit of patients in order to transcend the usual empirical medical practice and the new systematic medical practice. This new practice will allow the development of cutting-edge medicine, based on modeling, simulation, artificial intelligence, robotics and intelligent instrumentation techniques with the aim of having efficient healthcare for everyone and everywhere.

Many advancements have been made in healthcare in conjunction with technological development, both in hardware and software, that already reached a certain maturity. Also, the introduction of P4 medicine (preventive, predictive, personalized and participatory) has raised many societal expectations.

Therefore, scientists (doctors, mathematicians, physicists, computer scientists, etc.) must work hand in hand to build an interdisciplinary setting in the service of patients and to bring the power of AI to the patient’s bed.

Several countries are making this theme their priority and it is part of their innovation strategy. In France, four 3IA institutes were created and all of them make healthcare one of their strategic axes, as stated in the report by Cédric Villani (French mathematician and politician). The recovery plan presented by the government through the program “Investing for the Future” (PIA4), also presents digital technology, in particular healthcare, as a major pillar. The National Institute of Health and Medical Research (INSERM) makes AI in healthcare as a research strategy axis. Moreover,  the Ministry of Health created the Health Datahub to facilitate the use of health data in 2019.

In addition, several projects involving doctors, computer scientists, mathematicians and physicists are emerging in France, for instance, the project IHU Lyric of Bordeaux which is an institute of rhythmology and cardiac modeling. There is also the CONDOR project in Strasbourg, which revolves around the development of a control tower for operating theaters for real-time and automated control of surgical interventions. Also, the Paris-Santé campus project, which is a digital and AI institute for healthcare announced in December 2020 and will be part of a public-private partnership. Since 2018, we have been participating in this national dynamic by creating, at Aix-Marseille University, an institute in digital science and artificial intelligence for healthcare, in partnership with the APHM (Assistance Publique – Hôpitaux de Marseille) and the IPC (Institut Paoli-Calmettes which is a center for the fight against cancer). This institute will allow offering, in particular, a continuous value chain from basic research to medical application. Major universities like Oxford, Stanford, Berkeley, MIT and others are also making this topic of Artificial Intelligence applied to healthcare a strategic focus of their institutions.

This subject is showing such promise that the GAFAM (Giants of the Web – Google, Apple, Facebook, Amazon and Microsoft) are increasingly developing and funding research programs in this area, like the ALPHABET program developed by GOOGLE. We also note the position of industries in the electronics sector such as IBM’s Watson program, Intel’s SAFFRON program and pharmaceutical industries such as the Pfizer / Novartis / Sanofi alliance. We are therefore no longer in science fiction, but in an immediate reality on which our system must adapt to incubate innovation in AI according to the principles on which our health system is based.

In my opinion, nothing will be done without healthcare professionals. They are essential players in the success of this change. Their jobs will change profoundly and their initial and continuing training must be adapted as soon as possible in order to integrate this digital revolution. It is only by combining different skills and putting people at the center of our thinking that we will be able to achieve this transformation in the practice of medicine. This transformation, if carried out properly, could lead to better patient care, provide solutions to medical deserts, and be a source of very significant contributions in the exercises of professions in the healthcare.

2- Lte magazine:  AI depends on the existence of electronic data on patients. We know that there is currently a lack of such data in several countries. What do you suggest for Morocco to speed up the establishment of these useful databases so as not to miss the AI in healthcare shift?

Answer from Mr. Ouladsine: One of the major difficulties of research in AI for Healthcare is the access to real health or medical data. These health data are already the sinews of war, but also the black gold of AI in health and medicine. The various establishments, including universities, research organizations and healthc facilities, are all important producers of relevant health data (hospital care data, data from university and hospital preventive medicine, research data collected at the hospital or university, data from platforms in our research laboratories, etc.). However, what we are lacking are structures that facilitate their use by researchers. We believe that it is time for these different actors to join forces, at least at the level of each country, to give access to their data. This can be done by setting up one or more operational bases as well as platforms facilitating sharing and reuse in accordance with the rules of each concerned establishment. It will be necessary to make available to researchers corpus of health data, anonymized and respecting the principle of FAIRization of data (to make their reuse possible at any time). These corpora will consist of structured and unstructured data (as a reminder, in France, it is generally estimated that 80% of the data in a computerized patient record are textual) from current medical practice and/or specifically targeted studies. The anonymization of health data is an area of ​​research in itself that will need to be mobilized.

The emergence of these data warehouses will first require identifying the relevant data that could be reused (data from our various health research projects, health facilities, etc.) to ensure that this data meets security standards and current regulations regarding compliance with the General Data Protection Regulation (GDPR). Second, it will also be necessary to ensure interoperability by promoting the use of interoperable formats and syntactic and semantic standards to harmonize data in order to facilitate their processing, sharing and reuse.

An equally important point is that the creation of these research-oriented health data warehouses requires remarkable and modern resources in both hosting/storage and computing (access to high-performance computing). Thus, in France, a policy of developing Mesocentres (for computing) and Datacenter for data storage has been in place for several years.

Data in general and more particularly health data will represent the oil of tomorrow. The generation of health data warehouses must be part of the development strategy of many emerging countries like Morocco. If artificial intelligence is often considered a luxury, only accessible to Western powers, it represents a vector of development for others in my opinion. Since Morocco is a country with a strong rural component suffering from medical deserts, the use of digital technology and AI, connected medical devices and telemedicine represent a great opportunity.

3- Lte magazine:  There are already some health solutions based on AI and others are being developed. Should Morocco adapt them or should Morocco design new ones that take into consideration its own specificities?

Answer from Mr. Ouladsine: The applications of AI in healthcare are diverse and numerous. At CES in Las Vegas in 2018 (the world’s meeting place for consumer technology innovations), there was a lot of talk about artificial intelligence in the service of health. Virtual assistants are integrated into the daily lives of patients and connected objects are positioned to strengthen their prediction and analysis capacities. Thus, artificial intelligence allowed the development of smart medical objects that facilitate diagnosis (Two French startups have given life to “Motio”, a smartwatch that fights against sleep apnea, which affects 6% of the world’s population). Artificial intelligence is a valuable tool for preventing and tracking epidemics. Tools already exist for predictive medicine (predict kidney damage 48 hours before it occurs? This is possible thanks to DeepMind, which predicts Alzheimer’s disease by analyzing brain images or a blood sample, predict heart attacks based on an electrocardiogram (ECG), etc.). AI is also used in the development of new drugs (by studying billions of molecules, artificial intelligence can predict which ones correspond to a cell receptor or a virus). Implants and smart devices already exist (a knee prosthesis that monitors biomechanical constraints, a bionic contact lens that assists the blind, a robotic hand controlled by thought, a connected bandage that detects and prevents infections, smart T-shirts to measure heart rate and respiratory data, etc). The question that will arise is how to insert all these existing tools for better care practice.

Skills and knowledge exist in Morocco, AI is also developing in Morocco in various fields and research projects are supported like the Al Khawarizm program launched in 2019. However, in the field of health, I think that Morocco must first work to put in place processes to sensitize doctors of today and tomorrow to the use of digital technology and AI. One of the processes could be as part of their training. Indeed, to my knowledge, there is no initial training dedicated to digital sciences and AI for medical students. However, tomorrow, these doctors will have to use all kinds of digital tools including AI and will not be armed to exercise their expertise in an informed and critical manner. The generation currently in training was born with computers, the internet and smartphones. The goal is to give them the necessary bases to understand the tools of tomorrow. We believe that it is essential to start offering training in digital sciences and AI in healthcare.

Answer from Mr. Delliaux: To imitate and reproduce what is already in place is smart and saves a huge amount of time, but it has never allowed being ahead or independent. Alternatively, imagining, creating, building your vision and carving out your own path will always allow you to be ahead and autonomous. The great thing about digital and AI is that it does not require a lot of investments. It is not comparable to building nuclear power plants or developing a successful tourism industry that requires heavy infrastructures. Digital and AI are mainly intellectual work, and as such, any country is capable of harnessing the human genius of its population to contribute to it. Of course, Morocco can choose to put in place proven solutions elsewhere (in the United States, in China, in France?), But it can also choose to diagnose its specific needs and find internal resources to offer targeted treatments. It is for Morocco to make that choice, knowing that an ambitious AI policy for health systems is possible.

4- Lte magazine: Telemedicine has been around for a very long time but has grown dramatically, particularly with the imposed lockdown to face coronavirus pandemic. How can AI improve teleconsultation and telemedicine to make them more useful and practical?

Answer from Mr. Delliaux: Telemedicine and teleconsultation are not new. it was already performed between healthcare professionals where doctors used to call colleagues for a piece of advice. However, what has really changed is the democratization of this approach which only connects a health professional and a patient instead of two health professionals. This has been made possible thanks to technological developments: communication technologies, network infrastructures, data throughput, applications, user interfaces ergonomics, etc. Today most doctors, even patients, can be in digital contact if they have a smartphone. In 2018, the unique mobile users were 5.1 billion worldwide, 67% of whom were smartphone users. More interestingly, the patients of tomorrow will all come from the generation born with a smartphone in their hands which will increase those numbers. The acceleration in the use of teleconsultation and telemedicine that we have seen since the COVID-19 pandemic is just the beginning. So how will AI, another strong marker of technological developments in this decade and decades to come, be able to improve teleconsultation and telemedicine? We see at least two potential major contributions.

First of all, the first contribution concerns the classic medicine known as 4P: personalized, predictive, preventive, and participatory. Artificial intelligence and digital sciences are generally intended to optimize the personalization of patient care, i.e. to be as specific as possible for a given individual and to go beyond the group decision approach in medical decision making. In addition, the proliferation of sensors, and the facilities for collecting and transferring medically relevant information are an opportunity to detect by leveraging AI medical events at an infra-clinical stage in the context of prediction and prevention. In the digital age and even in the 19th century, the old proverb “Prevention is better than cure” remains particularly relevant. This approach can only be done with the patient’s consent, of course, but also and above all with their participation. Once data is acquired, monitoring the state of health of individuals is confronted with a bottleneck, the huge amount of data to be processed by doctors which are facilitated with AI. The second major contribution of AI in telemedicine/teleconsultation is filtering. In order to limit the inevitably scarce and precious expert medical resources, AI must filter and select only cases requiring medical advice. Once the acquired data is automatically processed, AI ​​will make it possible to assign patients to the appropriate resources. Digital technology and AI are not intended to replace humans but to provide them with support in everyday work. Since the human factor is essential to medical practice.

5- Lte magazine: AI applications are evolving rapidly on the one hand and on the other hand the chain connecting the patient to the doctor is expanding to other actors such as the developers of algorithms for example. How should we envisage collaboration between politicians and researchers for the design of new ethics in this area? Should we consider another “Hippocratic” oath for all the actors involved in this chain?

Answer from Mr. Delliaux: I don’t think we need another or a new Hippocratic Oath. It is sufficiently and inherently powerful and robust. And it is only opposable to the doctor. he ultimately remains the actor who makes medical decisions and acts. Rather, we believe that it is essential that doctors be trained for this technological change which, let’s say it, overwhelms or will overwhelm the medical profession. We strive for that. In addition, a hardware or software device for medical purposes which claims to be such and ready to be placed on the market must first obtain a CE medical device marking for any commercial use. This marking is the guarantee of a rendered medical service and therefore indirectly the Hippocratic Oath of the companies that develop them. Artificial intelligence can make no exception to this rule. Finally, another ethical safeguard is that of finances, because as everyone knows, money is the sinews of any war, be it against disease. Without money, there is no medical research, no health industries, and ultimately no health policy. A clear incentive for the development of tools including software adopting AI is to finance only ethically reasonable projects. Only benevolent research and development that puts the patient and society at the heart of the developed project, should be funded. It is up to funders to take sides and get involved. Some do it for ecology, others for sustainable development, it must also be done for ethical digital and AI in health.

It is with this vision that we develop all of our projects in close collaboration with representatives of users on the one hand (patient associations) and with representatives of the human and social sciences, philosophers and in particular ethicists. Let’s not lie to ourselves Healthcare is a business, It is up to us future users, as patients, but also and above all as health professionals to be demanding and benevolent.

Answer from Mr. Ouladsine: For the use of AI to be part of sustainable activity, people must be at the heart of all thinking. The accelerated arrival of AI and digital technology in the healthcare sector raises ethical issues mainly related to the use of sensitive data that ends up stored in large warehouses, hosted in the virtual space of a Cloud. This data should be used by researchers other than those who initially collected it in order to obtain the maximum information from it.  By allowing this, there is a risk of abuse and the recording of this data which poses a threat to the medical confidentiality reserved for the doctor-patient relationship. Artificial intelligence must be seen as a complementary tool necessary to contribute to the improvement of the medical practice of tomorrow by replacing humans on repetitive tasks. AI ​​analyzes a disease using only the characterizing data, however, the doctor does not only treat a patient according to his data but he treats him in all and different dimensions (emotion, feeling, etc) and. Since humans cannot be considered as the summation of their data, AI will never replace the human role of the doctor.

6- Lte magazine: How will the implementation of 4P medicine be considered while respecting the protection of patients’ personal data?

Answer from Mr. Delliaux: The question of the protection of personal data in particular that health data is, in my view, a false problem. Doctors have always been faced with medical confidentiality. It is even mandatory, it is the legal responsibility of the doctor to ensure that medical confidentiality is respected, including by those working with him. Such confidentiality concerns are not going to change tomorrow, even with the emergence of digital tools into everyday clinical practice.

However, we are faced with misuse of digital tools which consists of the use of smartphones or electronic mails to communicate between patients. It concerns the doctor as much as the patient. Who has not received or sent messages with medical content using public e-mails? Such as sending photos of medical exams, serology results, etc especially during the COVID-19 pandemic. The doctor is responsible for ensuring the strict maintenance of medical confidentiality. The only solution, in my point of view, is to make sure that the standards required by the regulatory and legal authorities in this area are respected, even if this adds complexity to the patient-doctor relationship.

 Policies should require developers to place safe products on the market through opposable standards and certified certification bodies. It is up to the law of the market to sort out the good solutions and the less good ones. Once again, benevolence and common sense should prevail over alarmist rhetoric. Health data is the new oil and thus the object of data theft, health data is sensitive and may be misused. The real issue is not the sensitivity or protection of health data, but the willingness and motivation of those who want to use it.

This interview was conducted in French by Ahmed Khaouja