Bio-inspired Computational Intelligence
The ODx Lab builds on new ideas to solve problems in a variety of situations from spectral analysis, to image and speech processing, to ECG analysis. The claim that a common processing algorithm is used by the brain to process both visual and auditory information has been backed up by biological experiments but very few actual algorithmic implementations exist. This model is exploited by converting the data to an invariant 2-d representation which would make processing independent of the data type. In addition, the notion of decisions based on ranking rather that absolute value (Rank Order Coding), which appears to be prevalent in the brain, is exploited. The proposed methodology will, therefore, introduce the advantages of a biologically-inspired process and additionally aid in supporting the claim of a common processing algorithm.
Computational Intelligence for Precision Medicine
Precision medicine requires developing the technologies needed to work with highly complex data sets and facilitate validation and development of diagnostics.
Converting Data into Knowledge
While there is a diverse array of approaches being implemented to convert data into actionable findings, they are all concerned with transforming complex heterogeneous data into a simple, comprehensible output, i.e. converting data to knowledge. As more data become available, there will be increasing need for models of increasing complexity to assist in analysis moving from molecular to cellular to physiological and finally to behavioural linkages. One necessary set of tools are systems that automate semantic harmonization and annotation for data and knowledge integration. Current methods remain difficult to use, mostly relying on human annotation. Another need is the adoption of standardized visualization which will aid the standardization required for clinical implementation and enable more rapid cross-model understanding and integration.
The Value of GUI
Currently large volumes of data remain underutilized and unexplored. One means of addressing the analysis bottleneck is to train ever-increasing numbers of bioinformaticians. However, the creation of graphical user interface (GUI)-based tools for users without coding knowledge would not only ease the bottleneck, but also greatly lower analysis costs.
Decision Support
The utility of precision medicine to daily clinical practice relies strongly on the availability of efficient tools to translate an individual’s data into diagnosis and targeted treatments. This is particularly true when it is expected that the user may not be computationally trained. Even in cases where analysis is performed by specialists, the analysis output must be designed with user-friendly principles in mind and be interpretable to the clinicians who will base treatment decisions on it. Models should strive to take account of the totality of datapoints and distil them into a recommended course of action.
