We are a MedTech company trialing a new type of Artificial Intelligence (AI) in the analysis of Clinical Trial (CT) data. Our AI methodology was developed by GENUTEX partner, Prof. Jacek Marczyk and has shown to be a reliable predictor of physical events based on the increasing “Complexity” of data collected prior to, and associated with, an event. Our technology has successfully been used in the manufacturing, defence, automotive, anti-counterfeiting and aeronautics industries. GENUTEX has the exclusive rights to use this technology in the pharmaceutical sector.
There are two key benefits of using our Artificial Intelligence Data Analysis (AIDA) technology when analysing CT data:
1. AIDA has the ability to analyse linear and non-linear relationships within large, often incomplete, biometric data-sets collected during the testing phases of pharmaceutical development and to rank, in order of importance (i.e. interdependencies between variables), the effect of an experimental drug on each measured biometric variable being recorded; and
2. AIDA has the ability (provided the correct biometric variables are being monitored) to detect the early onset of a medical condition. We believe the use of Complexity analysis in prophylactic medicine would be most appropriate in the development of drugs relating to conditions such as epilepsy, migraine, asthma, or any other medical condition that might benefit from early administration of medication.
Our technology platform is designed to electronically to capture patient data from smart-device wearables such as a smart watch or cardiac strap. These variables can be collected by us using our App and uploaded onto our cloud-based MedTech platform, “Vaultex”, where they will be stored and managed using the very latest in private, permission-based, templated and cryptographically secure blockchain technology.
Key attributes of Vaultex:
a. Our proprietary Vaultex platform uses a private, permission-based, templated and cryptographically secure blockchain ecosystem Hybrid refers to public/private access portals.
b. Open-source decentralised database software designed to facilitate rapid data collection in a legally enforceable virtual-environment;
c. Certificate Authorities (i.e. “Smart contracts”) are used as the basis for trust on our private blockchain and therefore computational-based ledger encryption. This allows patient identities to be encrypted and protected by Intel SGX;
d. Nodes are arranged in an authenticated peer to peer network allowing direct communication with one another (i.e. no “gossip” protocol is used hence speeding up the process of entry);
e. Data is shared on a need-to-know basis on our private blockchain. Nodes provide the dependency-graph of any ledger entry being sent to another node on demand (i.e. there is no global broadcast of all transactions);
f. Nodes are backed by a relational database and data placed in the ledger can be queried using SQL.
Data analysis that automates analytical model building by identifying patterns within data and systematising decisions with minimal human intervention is called Machine Learning (ML) and is a branch of AI. The problem with ML is that it requires numerous examples in order to learn to recognise patterns, which first must be defined, and then make decisions based on those patterns (i.e. by creating mathematical models). While ML works well in simple scenarios such as recognising a face or handwriting, it fails when analysing complex, non-linear, often incomplete data sets that may contain insufficient information to build a reliable model.
GENUTEX's AI is based on Quantitative Complexity Management, or QCM, which is a new paradigm in Artificial Intelligence. The key feature of QCM is that it is able to recognise the existence of anomalies the first time it is confronted with them. It is also able to pinpoint their sources thanks to a technique known as “Complexity Profiling”. This is why we say that QCM is Artificial Intuition, and has the ability to identify the existence of a (complex) problem without being trained to do so. This eliminates completely the need to build a model.
QCM is already being used in medicine – mainly in Intensive Care and Cardiology – in order to process large amounts of data, and to identify the drivers of patient instability and fragility, as well as to deliver precious early warnings of all sorts of (systemic) anomalies, especially in situations of high complexity, in which conventional data analysis techniques are not applicable. QCM is model-free, i.e. it doesn’t require a math model, and it takes into account all sorts of non-linearities in data, which traditional methods cannot.
A Complexity Map reveals the true structure of data, and this is paramount towards gaining a deeper understanding of given problem. The more complex and broad a data-set is, the greater the benefits of QCM analysis. Today, we are able to process datasets with hundreds and even thousands of variables, and extract insights, new knowledge, and better identify risk factors, beyond what conventional techniques can accomplish. Most importantly, this is done without the need to first generate numerous examples. This is immensely important in that a training set contains only the information that has been fed into it. A different training set will produce a different model. As QCM is a model-free techniques, as since it doesn’t require training, it eliminates to a very large degree any bias.
At GENUTEX, we analyse linear and non-linear data using algorithms which transform the data into an image, replicating the processes of the visual cortex. We do this by pixelating the data into an image and then use visual imaging techniques to determine the existence of single or multi-dimensional relationships in biometric data.