Our explorations

Kaduceo continues to explore and invest in research to develop our expertise in health data and exploit its full potential.

Essential values in Kaduceo’s corporate project, research and exploration are fully integrated in our daily life and our projects.

European research project: H2020 REMEDIA

Kaduceo is a member of the European consortium engaged to study the impact of environmental exposure on respiratory diseases.

Trajectory and care pathway

Patient clustering - Process mining - Data visualization

Complex and valuable data

To meet the quality requirements of the French health care system, it is necessary to place the patient at the center of the analyses and to have a good knowledge of their care pathways. The French National Health Data System (SNDS) is a valuable source of information for studying care pathways. These data are complex and include many categories of care: hospitalizations, consultations, medical and biological procedures, treatments and medical devices...

Our research aims to make these data interpretable and to develop robust and efficient tools for the classification, modeling and visualization of care trajectories.

Patient profiling

The identification of groups of patients with similar management, of typical pathway "profiles", and of a synthetic representation of these pathways, offers the possibility of questioning the efficiency of management. And this makes it possible to designate the most appropriate pathways from a medical and economic point of view.

Modeling and prediction of emergency room flows

Time series - Deep learning - Memory models

Observations and issues

The current situation in emergency departments is particularly difficult. Their overcrowding compromises the quality of care, and the exhausted nursing staff is faced with a feeling of helplessness. The peaks of affluence are particularly pronounced during epidemic periods, whether seasonal (bronchilitis, influenza, grastro-enteritis) or occasional (covid-19). More generally, the aging of the population, the difficulty of access to a city doctor, unjustified visits, etc. only aggravate the situation.

Development of our own models

We develop AI models (RNN) to predict patient flows, their impact on the different services, and to anticipate peaks in activity in the event of an epidemic. A better visibility on the expected level of affluence gives the hospital the possibility to react earlier and avoid saturation of its services.

Explicability of models and prediction results

Project carried out in collaboration with the Institut de Recherche en Informatique de Toulouse

Medical Imaging - Ophthalmology

Classification - Segmentation - Interpretability

Artificial intelligence and the future

Artificial intelligence technologies for medical image analysis promise great progress in facilitating the diagnosis of diseases and the interpretation of images. And they are all the more relevant in ophthalmology, where imaging plays a major role.

We are currently working on AI models for retinal disease screening and automatic anomaly detection.

Supporting practitioners

The objective is to allow ophthalmologists to save time or to delegate the diagnostic task. We also propose a tool for image annotation that allows us to collect reliable data, interpreted by experts, and all the more appropriate for the training of our models.