ED Mathématiques et Informatique
The algebraic lattice isomorphism problem
by Guilhem MUREAU (IMB - Institut de Mathématiques de Bordeaux)
The defense will take place at 14h00 - Salle de conférence Institut de Mathématiques de Bordeaux, bâtiment A33, Collège Sciences et Technologies, 33400, Talence
in front of the jury composed of
- Alice PELLET-MARY - Chargée de recherche - CNRS, Université de Bordeaux - Directeur de these
- Renaud COULANGEON - Professeur des universités - Université de Bordeaux - CoDirecteur de these
- Guillaume HANROT - Professeur des universités - ENS de Lyon et Cryptolab - Rapporteur
- Markus KIRSCHMER - Enseignant-Chercheur - Université de Bielefeld - Rapporteur
- Aurel PAGE - Chargé de recherche - Inria, Université de Bordeaux - Examinateur
- Adeline ROUX-LANGLOIS - Directrice de recherche - CNRS, Université de Caen - Examinateur
The Lattice Isomorphism Problem (LIP) asks whether two Euclidean lattices are isometric, and, if so, asks to compute such an isometry. This problem has been studied for several decades from an algorithmic point of view. Known methods rely, in particular, on computing short vectors in both lattices, a well-known hard problem in lattice-based cryptography. This thesis is devoted to the cryptanalysis of an algebraic variant of LIP, called the module-Lattice Isomorphism Problem (module-LIP). In this problem, lattices are endowed with the structure of a module over the ring of integers of a number field. The security of the post-quantum signature scheme Hawk, a candidate in the NIST standardization process, is closely related to the hardness of module-LIP. Before this thesis, the state of the art for solving module-LIP essentially consisted in forgetting this module structure and applying general methods for LIP. This generic strategy does not make use of the symmetries and arithmetic constraints specific to module lattices. The goal of this thesis is precisely to take advantage of this structure. We show that, for certain families of parameters, module-LIP can be reformulated in terms of classical problems from algorithmic number theory. In some cases, this translation leads to efficient attacks, and shows that the algebraic structure can help solve the problem more efficiently than in the unstructured case. However, the parameters used in Hawk avoid these attacks: the results obtained therefore do not directly compromise the scheme, but help clarify its security. More generally, this thesis provides a systematic study of module-LIP and its variants, using tools from algorithmic number theory. It clarifies the role of the algebraic structure in the hardness of the problem, identifies vulnerable cases, and thereby contributes to the security analysis of Hawk and of other potential cryptographic constructions based on module-LIP.
ED Sociétés, Politique, Santé Publique
Pharmacoepidemiological assessment of outpatient drug use in France: use of national, regional, and practice-based health data sources.
by Nicolas ROUSSELOT (Bordeaux Population Health Research Center)
The defense will take place at 14h00 - Salle du module 2.4 Université de Bordeaux, Campus Carreire,146 rue Léo Saignat 33000 Bordeaux, France
in front of the jury composed of
- Antoine PARIENTE - Professeur des universités - praticien hospitalier - Université de Bordeaux - Directeur de these
- Julie DUPOUY - Professeure des universités - Université de Toulouse - Rapporteur
- Caroline VICTORRI-VIGNEAU - Professeure des universités - praticienne hospitalière - CHU de Nantes - Rapporteur
- Charles CAZANAVE - Professeur des universités - praticien hospitalier - Université de Bordeaux - Examinateur
In France, there is a major challenge regarding the quality of medication use, to ensure its effectiveness and reduce avoidable iatrogenesis. General practitioners (GPs) are key actors in outpatient drug prescribing, the determinants of which are multiple and intertwined. Numerous data sources are available in France to study and improve their practices, but they are sometimes underused because of difficulties in access, lack of funding, or recruitment difficulties. Many clinical practice questions remain unanswered, although observational studies based on underexploited and accessible data appeared useful and feasible. This thesis aimed to study the value, limitations and challenges of pharmacoepidemiological assessment of the use and misuse of outpatient medications through the use of national, regional, and practice-based health data sources. A first study, conducted using the Medic'AM health insurance database, aimed to assess the impact of the COVID-19 pandemic and the first lockdown on the use of systemic antibiotics in outpatient care in France. The COVID-19 pandemic was associated with a rapid decrease in outpatient antibiotic use, with the exception of antibiotics intended for chronic indications and azithromycin. Our observations, during an exceptional period, suggest that there remains considerable potential for change in antibiotic use in France. This work confirmed the value of the easily accessible national Medic'AM data for monitoring major changes in medication use. However, this database does not allow analysis of treatment indications or changes in medication misuse. A second study, conducted using regional health insurance databases, aimed to evaluate the impact of the requirement introduced in May 2021 that pregabalin be prescribed on secure prescription forms, on its use and misuse in Nouvelle-Aquitaine. This regulatory change led to a significant and rapid decrease in chronic use and, above all, in new use of pregabalin. This impact was even more marked for misuse. This work further demonstrated the value of secure prescription forms in limiting medication misuse. These databases provide exhaustive regional data for monitoring changes in medication misuse when indicators can be identified in reimbursement data. However, they remain limited by the lack of clinical data needed to analyze misuse related to inappropriate indications, occasional misuse, or contraindications not visible in reimbursement data. A third study, based on data collection involving residents observing consultations during their general practice internship, aimed to describe NSAID prescribing practices among GPs in the former Aquitaine region. NSAID prescribing was generally appropriate, but comorbidities and high-risk drug interactions still need to be better taken into account, as does information regarding hydration and self-medication. This kind of data collection makes it possible to analyze medication use and misuse in general practice in detail thanks to the richness of the clinical data, with good feasibility and reduced bias through the observation of routine practices. But these studies remain limited in terms of representativeness and statistical power. Overall, this work illustrated the potential contribution of these three data sources as well as their respective limitations. Their combined use and the collaborations established open up perspectives for research on outpatient medication use. This work also highlights the need for a large-scale database containing patients' clinical data in France, such as P4DP.