Ph.D. Student


2021 – present : Ph.D. degree in Public Health (option Biostatistics), Université de Bordeaux 

2016 – 2018 : Master of sience in Public Health (Méthodes en évaluation thérapeutique : Biostatistiques, épidémiologie clinique), Université Paris Diderot – Paris VII

2012 – 2016 : Bachelor of sience in Biology (Biologie cellulaire et physiology), Université Paris Diderot – Paris VII


10/2021 – present :  Ph.D. student in Public Health (option Biostatistics)

02/2021 – 09/2021 : Consultant biostatistician for AIXIAL (CRO) in Sèvres, France

01/2019 – 02/2021 : Biostatistician for A.R.CA.D Foundation (Aide et Recherche en CAncérologie Digestive) in Levallois-Perret, France

08/2018 – 12/2018 :  Research engineer for IAME U1137, BIPID team, in Paris 18, France



[1] O. Margalit, B. Boursi , M. Rakez , A. Thierry, G. Yothers , N. Wolmark , D.G. Haller, H.J. Schmoll , Q. Shi, E. Shacham Shmueli , & A. de Gramont ; Benefit of Oxaliplatin in Stage III Colon Cancer According to IDEA Risk Groups Findings From the ACCENT Database of 4934 Patients. Clin. Colorectal Cancer. 2021. doi :

[2] B. Chibaudel , J. Henriques, M. Rakez , B. Brenner, T.W. Kim, M. Martinez Villacampa , J. Gallego Plazas, A. Cervantes, K. Shim, D. Jonker, V. Guerin Meyer, L. Mineur , C. Banzi , A. Dewdney, T. Sirisinha , H.J. Bloemendal , A. Roth, P. Thompson, M. Moehler , E. Aranda, E. Van Cutsem , J. Tabernero , H.J. Schmoll , P. M Hoff, T. André, & A. de Gramont Association of bevacizumab plus oxaliplatin based chemotherapy with disease free survival and overall survival in patients with stage II colon cancer: a secondary analysis of the AVANT adjuvant randomized trial. JAMA Netw . Open. 2020;3(10):e2020425 . doi :

[3] T. André, D. Vernerey , S. A. Im , G. Bodoky , R. Buzzoni , S. Reingold , F. Rivera, J. McKendrick, W. Scheithauer , G. Ravit , G. Fountzilas , W. P. Yong, R. Isaacs, P. Österlund , J. T. Liang, G. J. Creemers , M. Rakez , E. Van Cutsem , D. Cunningham, J. Tabernero , & A. de Gramont ; Bevacizumab as adjuvant treatment of colon cancer: updated results from the S AVANT phase III study by the GERCOR Group. Ann Oncol. 2020. PMID:31959341 doi

Thesis Title

 Modeling breast cancer screening and its impact on mortality reduction: Improvement with joint modeling and deep neural networks


Mammography screening programs are conducted to reduce breast cancer (BC) mortality by promoting earlier detection. Lead-time bias is known as the inflation of survival times in screen-detected patients due to cancer diagnosis brought forward by screening. We propose evaluating the lead-time bias using new statistical methods associated with joint modeling that considers some tumor features. The lead time correction will be done by subtracting the lead time estimated for each patient from her observed survival time. Simulation analyses will be used to assess the joint model’s performance, followed by real-world data from the Gironde general cancer registry.

Besides, dense breast tissue is a known independent risk factor for BC development and is commonly used to stratify women for supplemental screening examination. We will use a deep convolutional neural network to classify mammography images, and more specifically, assess breast density at the same level as experienced radiologists. We will train, validate and test our model by measuring the model’s agreement with the original radiologist based on the Breast Imaging Reporting and Data System categories.

Also, BC overdiagnosis due to early detection without survival benefit could be harmful and lead to over-treatment. To promote BC screening tailored to women’s risks and preferences, we want to make individual dynamic predictions of patient risk of cancer diagnosis and select the optimal time point to plan the subsequent mammography. We will use a joint model for successive mammography screening visits as recurrent event times and time to diagnosis. We will calculate, at each visit, a patient-specific cumulative risk for cancer diagnosis to decide for mammography time.

These newly created models will be available in user-friendly R packages to share with the community. Improving BC screening will limit early detection and orientate the focus toward advanced-stage disease, which fits better patients’ expectations.

Département de recherche santé publique Université de Bordeaux logo_ISPED logo_INRIA logo_HONcode