ALTRAN Spain is a company belonging to the Altran Group. Specialized in engineering services and R&D (ER&D), it offers to its clients a new way of innovating by developing the products and services of tomorrow. ALTRAN works with its clients in every part of the value chain of projects, from conceptualization to industrialization.
For more than thirty years, the group has provided its experience to key players in industries such as aerospace, automotive, defense, energy, finance, health sciences, rail and telecommunications, among others. In 2019, ALTRAN Spain generated revenues of more than 260 million euros, with a workforce of more than 4,500 employees.
In this emergency situation, Artificial Intelligence has shown its potential both for the fast diagnosis of the disease (eg.:, comparative analysis of radiographs or tomographies, chatbots for self-diagnosis), protecting the population (eg.:, analysis of thermal images), understanding the extent of the disease (eg.: big data to analyze population movements) or strengthen the public health system (eg.: predictive models to estimate the probability of survival or admission and length of stay in the ICU, as well as the effectiveness of the different treatments).
However, it has also shown some shortcomings related to the availability of the data necessary for AI algorithms to be trained, the necessary protection of the privacy of patients and their personal data, the need to establish agile mechanisms to promote public / private collaboration both national and international or the visibility of the mechanisms used by the learning algorithms to deliver their results.
In this context, we propose the development of a Federated and Interpretable Multipurpose Artificial Intelligence platform for rapid response to pandemics (AIMFIE) capable of supporting and allowing the combination of a series of learning models based on three fundamental pillars:
1. Healthcare and diagnostic support: This pillar is supported by machine vision techniques and predictive AI models.
2. Public health: This pillar is based on models to generate knowledge graphs to support, for example, semantic search engines or triage chatbots.
3. Secure and private collaboration: This pillar is based on the federated learning platform.
Among the main features of this platform, the following stand out:
Federated to allow the collaboration of different organizations, overcoming the barriers imposed by the necessary protection of the patient's personal data.
6/80
Secure by incorporating mechanisms to guarantee the availability and protection of data at the local level and avoid the exchange of data between different organizations thanks to the federated model.
Reliable by adding explanatory mechanisms (XAI, eXplainable Artificial Intelligence) to offer details on the processes followed by the algorithms to offer their results, thus favoring the understanding of the disease and the confidence of clinical staff.
Extensible to allow the incorporation of new learning models.
Open to allow third parties to exploit the models supported by the platform.
Scalable to support a growing number of federated organizations.
El principal objetivo del proyecto es la creación de una plataforma de servicios distribuida, abierta, escalable y extensible basada en Inteligencia Artificial multipropósito para la respuesta rápida frente a la aparición de pandemias.
All rights reserved
/
Search inscriptions
