The research theme on AI and healthcare aims at applying some of the IIIA AI techniques to the field of healthcare. Specifically, it is focused on the design of novel algorithms to provide solutions able to incorporate advanced Descriptive, Diagnostic, Predictive, and Prescriptive capabilities to Clinical Decision Support Systems (CDSS).
Contact: Eva Armengol
The current trend of moving towards a more Predictive, Preventive, Personalized, and Participatory medicine, known as 4P Medicine, is changing the healthcare paradigm. Digital technologies are playing an important role in this 4P paradigm generating a volume and variety of information never seen before. Artificial Intelligence is contributing by providing tools for the management and exploitation of this huge amount of data.
The prescription of highly personalized treatments increases the complexity of the knowedge and decisions to be considered. Artificial Intelligence and Machine Learning aims at developing innnovative decision support systems to speed up the discovery and consolidation of new evidence.
Time Series Analysis
Many information sources in healthcare have a temporal dimension. The explosion of biometrical sensors and wearables is an example of its relevance and an also of its noisy nature. Providing robust and efficient algorithms to deal with this amount of data is a challenging problem we are currently focused on.
Deep Learning
Deep learning allows computational models that are composed of multiple processing layers to learn representations of data with multiple levels of abstraction. These methods have dramatically improved the state-of-the-art in speech recognition, visual object recognition, object detection and many other domains such as drug discovery and genomics.
Case-Based Reasoning
CBR systems are capable of solving new problems using domain knowledge and the experience acquired in solving precedent problems (cases). CBR is a powerful methodology that allows incremental prototyping and short design cycles. Our group is an international referent on CBR with high impact contributions both in research and in applications.
Probabilistic Graphical Models
Probabilistic graphical models are a powerful framework for representing complex domains using probability distributions, with numerous applications in machine learning, computer vision, natural language processing and computational biology. Graphical models bring together graph theory and probability theory, and provide a flexible framework for modeling large collections of random variables with complex interactions
Causal Reasoning
Distinguishing between co-occurence and causality is one of the main challenges in healthcare. Determining causal relationships and designing robust causal models from data usually requires of the combination of multiple and heterogeneous data sources. Our research has been exploited into technology transfer projects.
Semi-Supervised Learning
One of the main characteristics of healthcare datasets is that they are usually partially annotated. Annotating and curating information is one of the key issues to obtain high quality datasets. This task requires a titanic effort and easily it becomes unaffordable. Semi-supervised techniques focus on minimizing the amount of labeled information, i.e. expert resources, while maximizing the models generated.
Explainability,
Trust, and Accountability
The adoption of complex AI/ML algorithms to take critical decisions collides with the requirement to understand Why these systems are recommending their decisions, which is their robustness, and the ethical consequences of these decisions. These systems will not succeed in healthcase if they do not incorporate explainability capabilities.
