This theme has been chosen because of its promising emergence. The use of algorithms is part of tomorrow's medicine, making it more personalized, predictive, participative and preventive. In the field of prevention, research projects, experiments and initiatives have been launched in areas as varied as the onset of epidemic crises, personalized monitoring of at-risk individuals and improving individual behavior to delay dependency. Promising avenues are emerging at both individual and local levels, as well as nationally and internationally. At the same time, as the subject is relatively recent, a number of questions remain open. These include the choice of models and the type of data collected. We are also beginning to see the impact of such developments on insurance coverage, business models and, more generally, lifestyles. These impacts deserve to be explored.

Through a series of seminars, the Polytechnique-Santé group proposes to explore the following question: Will preventive healthcare be algorithmic?

A series of five seminars is planned to address the subject (four others are in reserve). Each deals with a different facet of the theme. Cross-cutting questions will guide the discussions. At first glance, they concern: the models themselves (data and their processing).

Program

    Personalized follow-up of a person at risk (May 25, 7pm, Maison des Polytechniciens)

The processing of genomic and biomarker data, combined with other data, now makes it possible to establish the risk of a disease occurring at a very early stage. A number of initiatives are underway to ensure personalized follow-up. But questions are being asked about the length of this follow-up, the associated economic model, and the effects on privacy. The personalized approach also requires the collection of both clinical and behavioral data. Taking on multimodal data collection approaches is a challenge in itself.

Speakers: Fabrice André & Suzette Delaloge (Interception Program, Gustave Roussy)

Discussants: Emmanuel Bachy (Lyon), Alexis Hernot (Calmedica)

2.    Risk prevention and its impact on health insurance coverage (June 28, Académie de Médecine)
Knowledge of the risk of disease occurrence, thanks to predictive algorithmic methods, calls into question a founding principle of insurance coverage. If this risk is known, pooling it within an insurance system loses its meaning, and a move towards individualized contracts is a plausible option. At the same time, massive data processing makes it possible to monitor recommended behaviors using connected tools. With a connected watch, for example, it's possible to know whether a given policyholder is following recommendations on physical exercise or other matters. Contracts can be modulated according to compliance behavior. Prevention, both in terms of risk prediction and compliance with the measures implemented, therefore seems an additional dimension for insurers to integrate. But the adoption of such an approach depends on cultural contexts. Is it possible to envisage individual forms of insurance based on greater control of policyholder behavior in contexts such as France? If this seems more plausible in liberal contexts, isn't there a French-style model to be created to take account of these new forms of prevention in the insurance world? If so, what are its foundations?

Speaker : Eric Carreel (Whitings)
Discussants: Eric Chenut (Fédération Nationale de la Mutualité Française), Thierry Martel (Groupama)

3.    Prevention in crisis management (pandemics and others) and predictive models (Paris Santé Campus, October 3)
Mass data processing can help analyze the epidemiological evolution of a crisis between different geographical areas. The Covid-19 pandemic was a prime example of this. At the heart of these analyses lies the question of predictive reliability. Can the models be used as a basis for classic short-range forecasting, or should we restrict ourselves to nowcasting? At the heart of these epidemiological approaches are questions about the capacity to collect data according to the type of crisis, and the role of these models in initiating action.
Speaker: Antoine Flahaut (Global Health Institute, Geneva)
Discussants : Joëlle Barral (Google), Benjamin Garel (Covid crisis operator), Stéphanie Combes (Health Data hub)

4.    Can the quest for "ageing well" benefit from algorithmic approaches based on massive data?
Staying healthy and independent for as long as possible is a public health issue. Identifying bio-markers of aging, and combining them with other lifestyle data, seems a promising approach to meeting this objective. A better understanding of the biological and behavioral processes involved could lead to the implementation of preventive programs to avert dependency. The aim of this seminar is to take stock of a recent initiative in this field, the Inspire program, and to draw lessons for actions that could be carried out on a large scale. What do we know so far? What lessons can we draw for prevention?
Speaker: Bruno Vellas (Inspire program coordinator, Toulouse)
Discussant(s): Olivier Guérin (Geriatrician, Nice), others to be defined

[1] Etienne Minvielle et Hervé Dumez (i3-Centre de Recherche en gestion) in collaboration with the following members: Jean-Marc Aubert, Emmanuel Bachy, Joelle Barral, Juline Billiet, Julie Chabroux, Stéphanie Combes, Guillaume Couillard, Benjamin Garel, Karim Hatem, Alexis Hernot, Eric Labaye, Thierry Martel, Dorothée Moisy, Jérôme Nouzarède, Patrick Olivier, Ségolène Perrin, Philippe Peyré, Dominique Rossin, Arnaud Vanneste