Advanced data mining to enhance medical decisions based on complex, heterogeneous and scarce health data.
Learn about our data science expertise in precision medicine:
- What challenges do we address?
- What technology do we master?
- What do we publish about our projects and the projects of our customers?
Health care is facing several challenges:
- Several medical indications are still nowadays subject to wrong and/or inefficient diagnoses and therapeutic strategy choices
- The industry is subject to difficulties to come with new efficient therapeutics.
- Medical technologies have huge societal costs.
- An ever increasing pressure is put on budgets from health payers.
- Patients are more and more empowered and request for more and more tailoring in the management of their health.
- In all fields of health care, from fundamental research to market surveys, we observe a data deluge.
In that context, DNAlytics offers a tailored and on-demand consultancy service in data mining. By making use of Artificial Intelligence techniques, we build new decision-support tools from public and private datasets. As an introduction to what we do, we suggest watching the short video at the top of this page.
Our data mining platform fits with many data production technologies as well as various medical areas and needs. Although perhaps not explicit from the start, most of the project we are confronted with pursue three objectives:
- Build and validate predictive solutions
- Identify a small set of relevant (bio)markers from a vast set of possible markers
- Prototype software applications making those solutions available to practitioners
Through most of our projects, we use software elements that combine publicly available code and our own code library. A part of our website is dedicated to relevant software elements developed and/or maintained at DNAlytics. We also invite you to read about some actual projects we have performed in the past.
About Big Data: In most cases, when the GAFAs (Google, Amazon, Facebook, Apple, …) discuss the concept of “big data”, theyr refer to their own situation where many observations (millions of users) are available, each of them however described by a limited set of features. In this context, statistical learning, i.e. Machine Learning, is quite an easy task. In medicine, it is in most cases the opposite: due to financial, ethical, logistical constraints, databases generally include a very limited number of observations (patients) with respect to the number of features describing each of them (tens of thousands of genes and other covariates). In that context, Machine Learning is much harder, and it is precisely the context for which DNAlytics developed its own, recognized, expertise. That, also, makes us unique.
Our core expertise is in data science. We build predictive models and identify what (combination of) markers (in a broad sense) should contribute to these models. These models enable making predictions (such as diagnosis, prognosis, treatment guidance, adverse event prediction, etc).
We are able to deal with very large datasets (many patients) and very broad datasets (many features). To manage and analyse data, we rely on our own code written in R and C languages (mainly), and on various open-source libraries.
To obtain the computing power we need, we make heavy use of cloud computing solutions, such as the Amazon Web Services (AWS). We are an AWS Consulting Partner.
We are able to deal with very large datasets of many different kinds: epigenetics (e.g. methylation), genetics (DNA), transcriptomics (mRNA, lncRNA, miRNA, …), proteomics (e.g. mass spec.), elisa, metabolomics, clinical, epidemiologic, psychological, demographic data.
Ion torrent, Affymetrix, Illumina, Taqman, …
On top of that, we can tap into publicly available datasets to complete the data provided by our customers.
We master more classical statistics too (hypothesis testing, statistical analysis plan design/writing/execution). In the context of the conduct of clinical studies, we can also deploy electronic data capture tools (EDC), such as OpenClinica.
Here is a list of publications (scientific communications, patents, software libraries).
- Development of a multivariate prediction model for nocturia, based on the most important etiologies of the urinary tract
Roggeman S., Olesen T.K., Denys M-A., Goessaert A-S., Bruneel E., Decalf V., Helleputte T., Gramme P., Everaert K., European Urology. Supplements Volume 17, Issue 2, March 2018, Page e347 (2018) – Customer : Ferring
- Long-term Outcomes with Anti-TNF Therapy and Accelerated Step-up in the Prospective Pediatric Belgian Crohn’s Disease Registry (BELCRO)
Lucas Wauters, Françoise Smets, Elisabeth De Greef, Patrick Bontems, Ilse Hoffman, Bruno Hauser, Philippe Alliet, Wim Arts, Harald Peeters, Stephanie Van Biervliet, Isabelle Paquot, Els Van de Vijver, Martine De Vos, Peter Bossuyt, Jean-François Rahier, Olivier Dewit, Tom Moreels, Denis Franchimont, Vincianne Muls, Fernand Fontaine, Edouard Louis, Jean-Charles Coche, Filip Baert, Jérôme Paul, Séverine Vermeire, Geneviève Veereman, Inflammatory Bowel Diseases (2017) – Customer : UCL/BELCRO
- Routine Use of Synovial Biopsies for Diagnosis and Treatment Guidance
Thibault Helleputte, Pierre Gramme, Annals of the Rheumatic Diseases (2016)
- Prediction of peanut-challenge outcome with biomarkers
Peillon A., Thébault C., Helleputte T., Gramme P., Agbotounou W.K., Martin L., Prof. Dupont C., Benhamou P.-H., H.A. Sampson, Ruban C., EAACI 2016 –Vienna, Austria (2016) – Customer : DBV Technologies
- Monitoring viaskin peanut treatment progression with a biomarker-based model
L. Salmun, A. Peillon, C. Thebault, C. Ruban, R. Lambert, T. Helleputte, P. Gramme, Annallergy, Volume 117, Issue 5, Supplement, Page S102 (2016) – Customer : DBV Technologies
- Evaluating risk of esophageal variceal bleed in children with cirrhosis and waitlisted for liver transplantation
Stephenne X., Bonnet N., Varma S., Helleputte T., Smets F., Veyckemans F., Eeckhoudt S., Hermans C., Sokal E., Journal of Pediatric Gastroenterology and Nutrition (2016) – Customer : Cliniques Universitaires St Luc, service of paediatrics – See the resulting web app : Online Calculator
- Type of treating physician is associated with long-term disease outcome in the prospective Belgian paediatric Crohn’s disease registry
L Wauters, F Smets, E De Greef, P Bontems, I Hoffman, B Hauser, P Alliet, W Arts, H Peeters, S Van Biervliet, I Paquot, E Van de Vijver, M De Vos, P Bossuyt, J-F ahier, O Dewit, T Moreels, D Franchimont, V Muls, F Fontaine, E Louis, J-C Coche, J Paul, F Baert, S Vermeire, G Veereman, Journal of Cronhs & Colitis (2016) – Customer : UCL/BELCRO
- Heterogeneity of Synovial Molecular Patterns in Patients with Arthritis
Bernard R. Lauwerys , Daniel Hernández-Lobato, Pierre Gramme, Julie Ducreux, Adrien Dessy, Isabelle Focant, Jérôme Ambroise, Bertrand Bearzatto, Adrien Nzeusseu Toukap, Benoît J. Van den Eynde, Dirk Elewaut, Jean-Luc Gala, Patrick Durez, Frédéric A. Houssiau, Thibault Helleputte, Pierre Dupont, PLoS One (2015)
- DNA Methylation-Guided Prediction of Clinical Failure in High-Risk Prostate Cancer
Kirill Litovkin, Aleyde Van Eynde, Steven Joniau, Evelyne Lerut, Annouschka Laenen, Thomas Gevaert, Olivier Gevaert, Martin Spahn, Burkhard Kneitz, Pierre Gramme, Thibault Helleputte, Sofie Isebaert, Karin Haustermans, Mathieu Bollen, PLoS One (2015) – Customer : FFMI/KULeuven
- Inferring statistically significant features from random forests
J. Paul and P. Dupont, Neurocomputing (2015)
- Clinical Parameters vs Cytokine Profiles as Predictive Markers of IgE-Mediated Allergy in Young Children
Catherine Lombard, Floriane André, Jérôme Paul, Catherine Wanty, Olivier Vosters, Pierre Bernard, Charles Pilette, Pierre Dupont, Etienne Sokal and Françoise Smets, PLoS ONE (2015)
- Kernel methods for heterogeneous feature selection
J. Paul, R. D’Ambrosio and P. Dupont, Neurocomputing (2015)
- A generic cycling hypoxia-derived prognostic gene signature: application to breast cancer profiling.
Boidot R., Branders S., Helleputte T., Rubio L.I., Dupont P. and Feron O., Oncotarget, Vol. 5, No. 16, 2014. (2014) – Customer : UCL/PHARMA
- Baseline measurements of Coll2-1 and Coll2-1NO2 in urine are highly predictive of joint space narrowing in knee osteoarthritis.
Henrotin Y., Kraus V., Huebner J., Helleputte T., Deberg M., EULAR Congress, Madrid, Spain, 2013. (2013) – Customer : Artialis
- RheumaKit, a new early diagnostic tool for patients with arthritis.
Helleputte T., Dessy A., Hernandez-Lobato D., Dupont P., Lauwerys B., Knowledge for Growth, Ghent, Belgium, 30 May 2013. (2013) – Winner of the Inspiring Young Scientist Award
- Assessment of risk of bleeding from esophageal varices during management of biliary atresia in children.
Stephenne X., Wanty C., Helleputte T., Smets F., Sokal E., 63rd Annual Meeting of the American Association for the Study of Liver Diseases (AASLD), Boston, Massachusetts, USA, November 9-13 2012 (2012) – Cliniques Universitaires St Luc, service of paediatrics – See the resulting web app : Online Calculator
- Assessment of risk of bleeding from esophageal varices during management of biliary atresia in children.
Wanty C., Helleputte T., Smets F., Sokal EM., Stephenne X., Journal of Pediatric Gastroenterology and Nutrition. December 20, 2012 (2012) – Customer : Cliniques Universitaires St Luc, service of paediatrics – See the resulting web app : Online Calculator
- Strategic Overview of Personalised Medicine.
Helleputte T., International Pharmaceutical Industry, Volume 4, No. 2, Spring 2012. (2012)
- Differences in the cytokine profiles of cord blood mononuclear cells from allergic and non-allergic infants
C. Lombard, F. André, C. Wanty, J. Paul, P. Dupont, E. Sokal, F. Smets, European Society for Paediatric Gastroenterology, Hepatology, and Nutrition (ESPGHAN’12), Stockholm, Sweden, April 27-28, 2012 (2012) – Customer : Cliniques Universitaires St Luc, service of paediatrics.
- Expectation Propagation for Bayesian Multi-task Feature Selection
Hernandez-Lobato, D., Hernandez-Lobato, J., Helleputte, T. and Dupont, P., European Conference on Machine Learning (ECML), Barcelona, Spain, September, 2010. (2010)
- Robust biomarker identification for cancer diagnosis with ensemble feature selection methods.
Abeel T., Helleputte T., Van de Peer Y. and Saeys Y., Bioinformatics Advance Access published on February 1, 2010. Bioinformatics, Volume 26, No. 3, pp. 392-398. (2010) – Partnering with Ugent/Y.Saeys – Supp Material
- Expectation Propagation for Bayesian Multi-task Feature Selection.
Hernández-Lobato D., Hernández-Lobato J.M., Helleputte T., Dupont P., Machine Learning and Knowledge Discovery in Databases. ECML/PKDD 2010. Lecture Notes in Computer Science, vol 6321. Springer, Berlin, Heidelberg (2010)
- Inductive Biases for Stable Feature Selection in High Dimensional Spaces: Applications to Gene Profiling and Diagnosis from DNA Microarrays.
Helleputte T., PhD Thesis, University of Louvain, September 2010. (2010) – Video
- Robust biomarker identification for cancer diagnosis using ensemble feature selection methods.
Abeel T., Helleputte T., Van de Peer Y., Dupont P., and Saeys Y., Third International Workshop on Machine Learning in Systems Biology (MLSB), pp. 135, Ljubljana, Slovenia, September 5-6, 2009. (2009) – Partnering with Ugent/Y.Saeys – Poster
- Biomarker Selection by Transfer Learning with Linear Regularized Models
Helleputte T. and Dupont P., Third International Workshop on Machine Learning in Systems Biology (MLSB), pp. 159-160, Ljubljana, Slovenia, September 5-6, 2009. (2009) – Poster
- Feature Selection by Transfer Learning with Linear Regularized Models
Helleputte T. and Dupont P., European Conference on Machine Learning (ECML), Bled, Slovenia, September 7-11, 2009. (2009) – Poster – Video
- Partially Supervised Feature Selection with Regularized Linear Models
Helleputte T. and Dupont P., 26th International Conference on Machine Learning (ICML), Montreal, Canada, June 14-18, 2009. (2009) – Poster – Video
- Clinical response to the MAGE-3 immunotherapeutic in metastatic melanoma patients is associated with a specific gene profile present prior to treatment
Louahed J., Gaulis S., Helleputte T., Dupont P., Gruselle O., Spatz A., Kruit Wim H J, Dreno B, Lehmann F, Brichard V, 33th European Society for Medical Oncology (ESMO) Congress, Stockholm, Sweden, September 12-16, 2008, 470129. (2008) – Customer : GSK
To which we contributed as (co-)inventors, or on which we have a license:
- Signature of cycling hypoxia and use thereof for the prognosis of cancer.
Feron O., Boidot R., Branders S., Dupont P., Helleputte T. (2015) – This patent has been invented in the context of a collaboration with Prof. Olivier Feron (UCL)
- Method and tools for predicting a pain response in a subject suffering from cancer-induced bone pain
Alvaro Pereira, Chantal Gossuin, Dominique Demolle, Denis Gossen, Thibault Helleputte (2014) -This patent has been invented in the context of a collaboration with our customer Tools4Patients
- Method for prediction of a placebo response in a individual suffering from or at risk to a pain disorder
Thibault Helleputte, Alvaro Pereira, Chantal Gossuin, Dominique Demolle (2014) – This patent has been invented in the context of a collaboration with our customer Tools4Patients.
- Method for classifying a cancer patient as responder or non-responder to immunotherapy
Dupont P., Gaulis S. and Helleputte T. (2010) – This patent has been invented in the context of a collaboration with GSK.
For more information about our software libraries, go to the dedicated page.