loader image

AI ethics in health: Transformations in medical decision-making in the AI era

Characteristics of the position:

Fonctions 

post-doctoral contract

Emploi type (referens III)

 

Catégorie

 

Corps

 

Quotité 

12 months / 100%

Assignment:

Chair “Ethics and AI”, MIAI and IPhiG, UGA

Context and work environment:

Structure description

  • IPhiG

The Institute of Philosophy of Grenoble (IPhiG) is one of the research units of the University of Grenoble Alpes. It is composed of 11 statutory members, from various components of the UGA, more than 20 doctoral students, and several post-doctoral students. It is a member of the regional doctoral school ED 487 “Philosophy: history, representation, creation”. Its activities are organized along three axes: “Practices: values, norms, institutions”, “Mind and cognition” & “History of philosophy”.

  • Chair “Ethics and AI”, MIAI

The “Ethics & AI” Chair is one of the chairs of the multidisciplinary Grenoble Institute of Artificial Intelligence (MIAI). It aims to develop over a period of 4 years (2019-2023) a philosophical understanding of Artificial Intelligence both through a conceptual and normative approach and by reference to various “terrains” of observation. It reflects the activities of the Institute of Philosophy of Grenoble (IPhiG), a research unit of the University of Grenoble Alpes. Its activities include meta-ethics, normative ethics, applied ethics, political philosophy, philosophy of law. This variety of points of view gives the Chair a potential for in-depth research on topics to which the relationship between ethics and AI is giving rise today. The scientific director of the chair, specialized in political philosophy, public ethics and philosophy of innovation, ensures the productive interaction of this variety of perspectives. The Chair contributes with 3 others to the “AI and Society” axis of MIAI, and maintains a close relationship with the research conducted on the other axes of the institute. Through the initiatives and projects it develops, the chair currently brings together 4 PhD students in philosophy and information and communication sciences; it also brings together 8 contributors, researchers from other disciplines, such as robotics and among the humanities and social sciences, clinical psychology, information and communication sciences, management sciences and marketing.

Team description (N+1 and colleagues) :  under the authority of …. Team composed of X agents (X A, X B, X C…)

The position offered is directly related to the research activity of the chairholder, as well as to that of his contributors and the 4 current PhD students.

Position’s mission and main activities  

Scientific argument :

The main objective of this one-year (12 months) postdoctoral contract is to contribute to answering a global question:

Are the various modalities of medical decision making likely to be fundamentally disrupted by the new informations produced by AI?

Some believe that medical decisions have been “denatured” (Kempf & Kempf 2016). Although IT and digital applications are at the corner of most diagnostic or care pathways, how they actually affect decision-making processes is not clearly assessed. Also, concretely, the extent of this potential “denaturation” is to date difficult to establish.

To address this issue, it is essential to first analyze the evolution of decision-making processes in light of :

– the evolution of society’s expectations of physicians

– the understanding that physicians have of their missions.

From this preliminary analysis, the initial global question will need to be broken down into several concrete questions in order to avoid theorizing at too great a distance from the pragmatic realities of medical practice. As examples of non-exhaustive questions:

  • If today the decision-making process is guided by the statistical results of cohorts of patients to which a singular patient is assimilated (evidence-based medicine), how will this change when algorithms have the power of reliable individual prediction? The cognitive process of assimilation leaves a vast field to the intuitive knowledge of the practitioner, which draws its legitimacy from the accumulated experience of successes and failures, but also from the empathy mobilized by the singularity of both the pathology and the affected person.

As a consequence of this change, where the cognitive process of assimilation disappears in favor of personalized predictive information supposedly more performative (4P Medicine), other questions arise:

  1. Does this change the notion of responsibility (Bourassa Forcier et al. 2020)?
  2. How is the notion of medical authority likely to evolve? Doesn’t the high performance and reliability of AI tools, both in data collection and processing, give them a form of authority over human judgment, which is always more fallible than the decision-making elements provided by AI?
  • The relationship between the physician and his patient is also undergoing changes. If the medical decision is part of a relationship between the doctor and the patient, we could say that, as a result of its increasing insertion in the phases of observing symptoms, understanding illnesses and analyzing treatment follow-up, AI is inviting itself into this relationship, but how does it modify the conditions? Is the trust necessary for the relationship already evolving on the part of patients, shifting from the human (the doctor) to the machine (Hatherley 2020)?
  1. Is it enough for algorithms to be “reliable” (Bernelin & Desmoulin-Canselier 2021) to take control of the decision? In other word, who tomorrow, will be recognized as competent to express the medical decision, the human or the machine?
  2. What could be the consequences of this algorithmic pre-eminence on the collective intelligence (multidisciplinary meeting) which is today the gold standard of difficult decisions?
  3. what impacts do these transformations have on the training of physicians and the health and care professions (Zeitoun & Ravaud 2019)?

Finally, to what extent are the ethical approaches that traditionally surrounded medical decision-making being modified? Do the terms that constitute this approach need to be redefined or reinvented today (Duguet et al. 2019)? And in these cases, according to which axes of reflection? For example, how should the medical relationship be managed in light of new forms of prediction (notably genomic, as these forms modify knowledge of the possibility of the onset of a disease) and given the multiplicity of tools available? What new partition between practitioner, machine, and patient (McDougall 2019)? How to operate through AI value choices in always specific clinical contexts (Tsafack Chetsa 2021)?

Through the study of cases in one or more areas of medicine and by monitoring emerging technological solutions as well as the literature in health AI ethics, the postdoctoral researcher will provide answers to these questions (and others that may emerge during the course of the research) by embedding their work in the ethical dimension in several ways:

1) by determining what new risks arise in decision-making situations with AI;

2) by illuminating the ways in which older forms of ethics are rendered obsolete by these situations;

3) by devising new ethical protocols that incorporate value choices in specific contexts.

Through his or her work in the ethics of AI in medicine, the postdoctoral researcher will collaborate with the members of the “Ethics&IA” Chair (its director, its permanent research contributors, doctoral students and postdoctoral fellows), and will contribute to the scientific animation of the Chair.

References

Bernelin Margo, Desmoulin-Canselier Sonia (2021) « Chapitre 2. L’intelligibilité des algorithmes dans les systèmes d’aide à la décision médicale », Journal international de bioéthique et d’éthique des sciences, 2021/2 (Vol. 32), p. 19-31. DOI : 10.3917/jibes.322.0019.

Bourassa Forcier Mélanie, Khoury Lara, Vézina Nathalie (2020) : « Liability issues for the use of artificial intelligence in health care in Canada: AI and medical decision-making », DMJ, Summer 2020, 46 (2), p. 7-11. DOI : 10.15273/dmj.Vol46No2.10140.

Duguet Julien, Chassang Gauthier, Béranger Jérôme (2019) : « Enjeux, répercussions et cadre éthique relatifs à l’Intelligence Artificielle en santé. Vers une Intelligence Artificielle éthique en médecine », Droit, Santé et Société, 2019/3 (N° 3), p. 30-39. DOI : 10.3917/dsso.064.0030

Hatherley Joshua James (2020), « Limits of trust in medical AI », Journal of Medical Ethics, 2020, 46, p. 478-481. DOI : 10.1136/medethics-2019-105935.

Kempf Aude, Kempf Emmanuelle (2016) : « L’informatisation de l’aide à la décision : la décision médicale est-elle indemne ? L’exemple d’un outil prédictif en cancérologie », Revue française d’éthique appliquée, 2016/1 (n° 1), p. 59-70. DOI : 10.3917/rfeap.001.0059.

McDougall, Rosalind J. (2019) : « Computer knows best? The need for value-flexibility in medical AI », J. Med Ethics, 45, p. 156-160. DOI:10.1136/medethics-2018-105118

Tsafack Chetsa, Ghislain Landry (2021) : « AI in the Medical Decision Context », in Towards Sustainable Artificial Intelligence. A Framework to Create Value and Understand Risk, Apress, Berkeley, CA., p.103-121. DOI : 10.1007/978-1-4842-7214-5_8

Zeitoun Jean-David, Ravaud Philippe (2019) : « L’intelligence artificielle et le métier de médecin », Les Tribunes de la santé, 2019/2 (N° 60), p. 31-35. DOI : 10.3917/seve1.060.0031.

Scientific disciplines: philosophy, sociology, computer science, robotics, applied mathematics, statistics, medical studies

Sub-disciplines/research areas (in no order of priority): ethics of care/health and bioethics, political philosophy, philosophy of techniques, ethics of AI (computer ethics/algorithmic ethics, artificial ethics, ethics of robotics, ethics of AI uses), social philosophy.

Keywords: ethics, AI, robotics, digital, health and care, patients, patient-experts, health institutions and care structures, doctor-patient relationships, social and emotional support of patients, body, innovation, autonomy, trust, compliance, public and private liberties, privacy, vulnerability.

Terms of the competition: Candidates are invited to propose to the Jury, with their Academic Curriculum, a personal research project (10 pages maximum), including a description of their future activity within the “Ethics & AI” Chair that falls within the strategic orientations of the latter. If the position concerns philosophy, the multidisciplinary aspects of the candidate’s profile and the multidisciplinary perspectives of his/her project will be highlighted, in dialogue with the disciplines of the contributors to the Chair.

The proposed project must be able to fit into the “AI for humans and the environment” axis of MIAI, and more particularly be able to create a sustained dialogue between axes 4 (“AI and society”) and 5 (“Health”).

Fluency in French and English (oral and written) is a prerequisite to apply, as the admission interview will be conducted in both languages.

Fluency in French and English (oral and written) is a prerequisite for applying, as the admission interview will be conducted in both languages.

Start date: September 1, 2021.

End of contract: August 31, 2022.

 

Missions / duties performed:

The purpose of the postdoctoral contract offered to the competition is:

– To contribute to the advancement of scientific knowledge in the field of AI ethics (as defined above) in health and care, including new conceptual and normative approaches, which are grounded in fields of study.

– To maintain the dialogue between philosophy, the SHS disciplines developed in the chair (social and clinical psychology, management sciences, information and communication sciences) and the scientific and technical disciplines developed within the MIAI institute (computer science, robotics, cognition), as well as with the world of health and care, whether academic (medical and pharmaceutical research laboratories), hospital (CHUGA) or other (patients, caregivers, associations, start-ups and other companies in the health and care sector).

– To feed the IPhiG’s axis 1 while possibly interacting with the two other axes.

Main activities (in order of importance):

  1. Publication in academic journals, especially in English language.
  2. Contribution to the scientific animation of the “Ethics & AI” chair: organization and intervention for conferences, seminars.
  3. Setting up scientific projects (ANR, European programs, etc.).
  4. Contribution to the outreach of the “Ethics & AI” Chair: oral intervention in front of non-academic audiences (health care personnel, patients, students, etc.).
  5. Invitation to participate occasionally in training sessions (doctoral training on the Grenoble site, MIAI continuing education).

Event – Objective result(s) by the end of the contract:

– At least 2 research articles in English (published or submitted) over the duration of the contract,

– Contribution to the organization and animation of the program “Etats généraux de l’éthique de l’AI”.

 

Evaluation and control of the achievement of the result(s):

Publication of the offer: October 31, 2021

Deadline for sending applications: November 26, midnight (Paris time)

Examination of applications and Eligibility Jury: around November 29: Results and invitation to the oral exam for eligible candidates

Admission interviews (15-minute interview): during the week of December 6

Results of the competition: December 15

Start date: January 1, 2022

End of contract: December 31, 2022

Restriction or constraints related to the position :

Physical presence at the University of Grenoble Alpes for the duration of the contract is a necessary condition, as the position to be filled includes the coordination of a scientific team as well as scientific animation functions on the campus and in the Grenoble area, and in the Auvergne-Rhône Alpes region. An explicit commitment from the candidates will be required during the recruitment interview.

Desired profile :

Expected skills (priority) :

Academic excellence

– Relevance of the research project to the strategic orientations of the “Ethics & AI” Chair

– International research experience: publications, participation in seminars and colloquia, stay in research units

– Ability to engage in multidisciplinary dialogue as demonstrated by previous activity

  • Trade skills/ expertise

Expression in French and English language, written and oral.

  • Personal skills

– Autonomy,

– Ability to lead a small scientific team,

– Ability to mobilize the right interlocutors in an academic and peri-academic environment (health and care fields, scientific, technical and industrial mediation fields).

Supervisory mission: ☐ Yes ☒ No
Number of agents supervised per category:      ….. A, ….. B,  …… C

Desired professional experience: ☐ beginner ☐ 2 to 5 years

Experience in setting up research projects (national and European) is a plus in the file.

Previous formation, diplomas:

PhD required in philosophy and another discipline developing AI ethics.

General information

The level of remuneration will be determined according to the seniority and skills of the candidate, and within the framework of the remuneration scale in force at the University Grenoble Alpes. Indication: between HS 01 and HS 03.

Contact for the questions related to the position:

First name, LAST NAME, position:

Prof. Thierry Ménissier

Head of the Chair « ethics & AI », MIAI/IPhiG

Mail : thierry.menissier@univ-grenoble-alpes.fr