Projects

Computational technologies for sense-making and uncertainty management allow relevant information to be identified, useful insights to be gained, and timely actions to be taken in complex decision problems. Many recent global events of significant and sometimes catastrophic impact, e.g.,business globalisation, disease outbreaks, terrorism attacks, earthquakes and tsunamis, etc., have highlighted the urgent need for decision support systems that can efficiently and systematically help manage uncertainties and minimise surprises.

The major challenges brought about by substantial contextual, informational and temporal changes in such decision situations are being addressed by the multi-disciplinary Biomedical Decision Engineering (BIDE) research group. BIDE is formed by faculty staff, research staff, and students from the Medical Computing Lab (under School of Computing), the Systems Modeling and Analysis Lab (SMAL) (under Department of Industrial and Systems Engineering, the Faculty of Engineering) and the Faculty of Medicine (Department of Medicine, Department of Paediatrics and Department of Diagnostic Radiology) at NUS. It also involves collaborators from various local and overseas universities and hospitals.

Our long-term research agenda aims to develop advanced computational technologies to support complex decision making in dynamic environments. An integrative approach based on Bayesian theory, decision theory, and cognitive theory serves as a unifying foundation for our work. The analyses are motivated by and the results evaluated in selected prototype applications in major domains where the new technologies would make significant impact: e.g., biomedicine, e-commerce, and homeland security.

Our current research focuses on developing a comprehensive decision making framework that:

supports adaptive reasoning in a changing, uncertain environments
integrates and learns from distributed information sources; and
provides timely decision recommendations with limited resources.

This framework includes:

languages for effective specification of decision factors and objectives in a context-sensitive manner,
methodologies for reasoning with and learning of decision information from human experts and online databases, and
techniques for adapting responses and managing surprises under resource constraints.

Our current research activities include the following projects:

1. Project ResEasy

This translational research project adopts a novel collaborative approach by facilitating a research team, an engineering team, and a clinical team to speed up trial implementation and productization of our previous research results. The objective is to test trial the effectiveness and feasibility of an open, adaptive workbench which implements the decision support applications based on a set of generic information management toolboxes that support integration, visualization, analysis, security and reporting of the relevant information. The toolboxes can be generalized to other diseases or conditions directly and be adapted and deployed in multiple sites simultaneously. The ResEasy project initially focuses on facilitating best practices in process management, outcome analysis, and guideline execution in prospective care of Asthma patients, and acute care of Acute Respiratory Distress Syndrome (ARDS) patients.

2. Decision Making with Multimodal Information

Projects in this area aims to combine multimodal information, e.g., text-based, structured, and/or image data, to support biomedical decision making. The objective of an initial project is to develop an intelligent human organ segmentation system using 3D medical magnetic resonance images (MRIs) to support medical decision making. We have proposed hybrid image processing algorithms and evaluated them on set of Kidney Images. We continue to explore and develop other Image Processing algorithms and decision support system models.

3. Decision Modelling in Dynamic and Uncertain Domains

Projects in this area develop new modelling and reasoning techniques that combine advances in hierarchical and logical knowledge representations with Bayesian network formalisms and stochastic processes. One focus is on extending the expressiveness, efficiency and interface of a general dynamic decision modelling framework, called DynaMoL (Dynamic Decision Modelling Language) that has been commercialised by the group. These modelling frameworks support automated and interactive formulation of decision models with factors that change over time and under uncertainty, with multi-level, multi-perspective, and multi-modal inference capabilities.

4. Machine Learning from Complex and Large Databases

Projects in this area survey and develop new techniques for automated acquisition of numerical parameters, structural constraints, and temporal patterns in large databases, either in the number of data items or in the number of variables. This study provides some insights into the feasibility of such tasks forsupporting dynamic decision modelling in various domains. It also illuminates some limiting constraints inherent in the available databases. In addition to the core analytic framework, domain-specific challenges are also addressed in the different types of databases, e.g., databases of human genome sequences, microarray gene expression profiles, metabolic pathways, and clinical profiles in biomedicine, bartering trade and exchange records in e-commerce, and terrorism risk and anomaly hypotheses in homeland security.

5. Decision Formulation from Multiple Knowledge Sources

Projects in this area develop techniques for the automated construction and analysis of dynamic decision models. Such tools effectively retrieve, translate and integrate information from multiple sources. These sources include expert opinions, text-based articles, structured electronic records, public repositories, specialised experiment datasets, as well as personal data, e.g., medical records and genomic expression profiles, to support various decision modelling tasks.

6. Advanced Techniques in Probabilistic Graphical Models

Projects in this area explore and develop analytical techniques in the realm of Probabilistic Graphical models and Influence Diagrams. We are currently working on context-aware probabilistic reasoning, multiple-level probabilistic game representation and knowledge discovery using Bayesian learning. The results are being evaluated in selected prototype applications in biomedicine and e-commerce.

7. Time Critical Dynamic Decision Modeling

Projects in this area investigate methodologies and build computer-based tools for managing complex decisions under limited resources. Such techniques take into account the dynamic nature of the problem, the uncertainties, the preferences of decision makers, as well as the time-criticality of the problem, so that the decision models being built are optimum in size to offer timely recommendations for effective actions.