The Next generation Machine Learning
Bantham RADAR is our latest generation bio-informatics engine utilising Machine Learning, specifically designed to ingest information from virtually any data source to deliver real-world insight to your past, present and future operations.
Bantham RADAR helps you understand and improve decision making for improved business outcomes using a data driven approach.
There is nothing artificial about our intelligence.
What makes us different?
There are many Machine Learning platforms available today, each claiming to answer the BIG questions, address the BIG issues and provide new and superior insight into everyday challenges, previously beyond the reach of conventional analytics. Bantham Radar is no different, but where Bantham differs from many competitors is that we will not blind you with science and above all, you do not need to be a data scientist to use or benefit from our platform.
Bantham Radar is for anyone who wishes to use data analytics to improve outcomes. Radar will:
Analyse what has happened in the past.
Predict what is likely to happen in the future.
Explain why it has happened.
Model solutions to deal with that predicted future.
Bantham Radar has been developed to explore real-world applications of AI and is a fully automated, AI-driven machine learning platform designed specifically for the healthcare and bio-science sectors. It is able to deploy ‘automated data science robots’ that operate at extremely high precision performance. These robots can evaluate mission critical data to produce predictive models that solve real world problems. It then optimises these models to look for patterns or configurations of parameters that human modellers may not even consider or have the patience to develop. The platform is data agnostic which means it has a very broad range of use cases.
The system is designed to focus on problems that are beyond human scale in dimension or complexity.
Typically, a real-world problem may have tens or even hundreds of millions of data points which must be analysed during the data modelling phase. In addition, there are tens of thousands of different machine learning models that may need to be considered before finalising the selection of the model. The range of permutations of these models and the inherent complexity within the data makes this a ‘beyond human scale’ problem. Bantham Radar automates this process and can generate what would normally take a human weeks or months to achieve in just minutes or hours. In addition, while humans naturally tend to have biases, Bantham Radar does not have any. It is also distinctive because it intuitively automates a number of the critical steps required in AI, which improves on the quality of the output and the speed of delivery.
The Importance of your Raw Data
Much is made of algorithms, their efficacy, their speed and their ability to cope with voluminous data beyond human scale, however for the ‘intelligence’ to work properly, it is vital that the integrity of the datasets ingested is as close to 100% as possible.
Data preparation may not be glamorous, however if your raw data isn’t clean, validated and consolidated in the right way, any analysis is rendered meaningless, and in some applications (i.e. healthcare), improper use might actually be dangerous. All too often users spend too little time on data preparation, hoping that the AI engine will somehow make sense of it all. It won’t because it can’t.
To optimise data integrity, Bantham operates a five-stage data preparation process, each combining to make your data as clean and fit for purpose as possible.
Initial audit to see what data is actually available.
Eliminate duplicates, errors and remove data that is not relevant.
Remove data of poor quality or convert unstructured data (such as natural language) into structured datasets
Begin to join different datasets.
Aggregate datasets into formats that can be readily ingested by the Bantham Radar platform.
Making the Most of What You Have
Many organisations still rely on legacy systems for their key operations.
These often have little, if any, analytical capability and consequently they act as little other than repositories for historical data which are simply too expensive or disruptive to replace. Bantham Radar will literally bring their content back to life, giving you present day analytics and explainable AI capabilities without having to replace core systems.
Application of Big Data in Healthcare
From the diagram below, it is clear that the potential applications for Machine Learning are many and far reaching.
A recent real-world application involving 2,500 brain scans enabled the identification of different types of brain tumours with an efficacy of 84%. The results were attained after just two hours of machine ‘training’ and were delivered in a matter of a few minutes once training was complete. A larger sample size would inevitably have delivered even better results, enabling further fine tuning of input parameters.
The benefits to patient and clinician alike are both obvious and significant:
• Fewer mis-diagnosis’
• Validation of prior diagnosis
• Quicker detection allowing earlier treatment and better patient outcomes
• Significant time savings and cost reductions