Case Studies of Emphron Informatics Past Projects
All Emphron case studies are concerned with delivering solutions to real business problems, using advanced statistical and computational techniques. Emphron's involvement has been key to strategic business decision making, based on large and complex data sets, often involving integration of data from multiple sources.
Orange City Council -- Waste Management
The objective was to estimate the probability of contaminating local apiaries with American Foul Brood disease, from a proposed new waste management facility. The project required a review of the environmental and behavioural factors affecting contamination with Foul Brood, and quantitative modelling of the probability of transmission based on these factors. Probability was expressed as the mean time between infections. The outcome was an understanding of the risks of contamination, leading to the development of new mitigation strategies.
Lumigenix -- Ancestry Analyses
Development of genomic analysis to profile consumer ancestry from SNP data. Corresponding maternal and paternal history are then displayed in an online application in the form of interactive maps and text descriptions.
Lumigenix -- Genomics using the Cloud
Development of analysis and visualisation tools for deploying large scale genomic data mining in a cloud-based distributed computing environment. Both backend offline analysis and online interactive visualisation, tabulation and search functionality have been provided for a genomics research company to support augmented pharmaceutical trials.
Air Quality Impacts of Aluminium Refining
The objective was to measure the impact of refinery operations on air quality in adjacent communities. Large and complex data sets from nearby monitoring stations were used to assess air quality, and to identify the determinants of air quality (both operational and meteorological). Sophisticated models of wind effects were used to identify the major foci of emissions. It was demonstrated that even under extreme circumstances, the air quality was well within environmental health guidelines.
Port of Melbourne -- Water Quality
Develop a framework for decision making about turbidity arising from the plume associated with dredging the Port of Melbourne shipping channel. The project involved integration of multiple lines of evidence and multiple data sets, and relating these information sources to environmental management objectives. The framework was also used to define a formal process for cessation of monitoring when turbidity levels had returned to background, after the cessation of dredging.
Port of Melbourne -- Fish Stocks
Review the monitoring program for and evaluate captured data for fish stock levels in Port Philip Bay to access the impact of dredging upon local fish stocks. Develop decision criteria to identify possible impacts on fish species.
Port of Melbourne -- Fairy Penguins
Develop a framework for monitoring the health and conditions of fairy penguins in Port Phillip bay. The framework was implemented as part of the environmental monitoring programme for the Port of Melbourne Channel Deepening project.
Validation of a New Wheat Quality Assessment
Experimental design and protocol development for studies to support marketing of a new technique to measure the quality of wheat. The new technique offers substantial cost, accuracy and precision benefits over current methods, but it was necessary to develop a formal process to demonstrate this. Objective verification of test performance was a crucial business need, and had to be completed in an environment of potentially hostile review.
Expert Witness Evaluation of Sources of Contamination
By a combination of mathematical modelling, statistical analysis and use of nuclear magnetic resonance data we were able to demonstrate that materials in a defendant's surgery were not responsible for contamination of a pharmaceutical product. The work involved a combination of advanced statistical analysis, modern analytical chemistry, and an ability to communicate clearly but authoritatively. The court accepted the evidence.
Ovarian Cancer -- Diagnostic Device
Development and validation of a multi-analyte signature (essentially a statistical description of the relationships between the markers and the disease status expressed as a probability) for the diagnosis of ovarian cancer. The new diagnostic signature outperforms current diagnostics in terms of sensitivity and specificity. This external data review and analysis was a key factor in the client's ability to licence the product. The diagnostic has been delivered internationally, and further large scale trials are under way.
Diagnostic -- Hosted Applications
Development and deployment of a SOAP server for analysis of cancer data. The clinical laboratory interrogates the SOAP server and receives a diagnosis based on the measured variables provided. A complete report of the diagnosis and risk assessment is provided in PDF form. Additionally, an online administration interface is provided for control of user, role and access privileges.
Sepsis Diagnostic
Development of a diagnostic for Sepsis, based on blood gene expression markers. The objective was to discover and optimise a marker panel that would distinguish between systemic inflammatory response syndrome without bacterial involvement, and full bacterial sepsis. The project involved data mining from large databases of gene expression data, and validation of diagnostic performance using advanced statistical methods. The client's diagnostic is currently being implemented in a major hospital, and negotiations with major international diagnostic companies are in progress.
Immuno-Histochemistry -- Automated Workflows
The objective was to provide a secure platform for management of data, and common analysis workflows for a biotech company developing a new proteomics platform. The business objectives were to allow rapid but reliable analysis of complex multidimensional data by non-specialists. The project delivered an online data-storage and analysis system for customer-specific laboratory data. An administration interface was provided to control user, role and access privileges. Individual users can select from amongst the uploaded data sets and perform a variety of analyses with publication-ready graphs and tabulations output.
Ore Trains -- Discrete Event System Modelling
The objective was to maximise throughput in an ore train and tipper system through discrete event simulation modelling under different management scenarios. The project involved data analysis, and development of a simulation model. The model was used to plan and schedule investment.
Optimisation of an Application-Specific Computer System
The objective was to optimise the design of multiprocessor distributed high performance computer systems. The project involved development of a system to perform optimisations, delivering the best balance across cost, performance and energy targets. The approach used involved evolutionary optimisation to search large solution spaces.