SCIENCE
Expertise – Process engineer Phil Green with Dr Jennifer Thompson , regulatory compliance services manager at ITS .
making plus language processing and translation .
For example , the HSE is leading on the Discovering Safety programme , which uses AI to gain insights from the HSE data sets and businesses can sign up to a regular newsletter and get involved .
Machine Learning ( ML ) methods are a sub-set of AI and enable computer systems to learn without explicit instruction by using algorithms and statistical models to analyse data sets and determine underlying patterns . The models developed can subsequently draw conclusions or make decisions about new data .
Deep learning ( DL ) is a sub-discipline of machine learning . Data sets are analysed to create hierarchies of insights called neural networks , which mirror how the human brain processes information and can independently establish correlations between data inputs and outputs .
A large neural network can have several hundreds of layers and millions of processing nodes , so it can be incredibly challenging to explain “ how ” decisions are made compared to ML algorithms . Due to this complexity and lack of transparency , extra safeguards , such as human supervision , are often required when such models are deployed in highrisk contexts , for example , those with the potential to impact patient safety , product quality and data integrity .
Could your business be doing more to explore AI , to automate routine tasks or gain new insights on your business and manufacturing processes ?
The benefits can be significant due to increased efficiency , reduced waste , better business process outcomes or
Experts have suggested that investment in AI may add an additional £ 630bn to the UK economy by 2035 .
better consistency and quality of product manufacture , which all lead to improvements in productivity overall .
The first step in implementing AI / ML is to review where benefit can be gained , and which challenges are best suited to an AI approach rather than a traditional approach . Having identified the business goal , you will need a detailed understanding of the business and / or manufacturing processes to proceed .
Guidance on AI model development and implementation for the life sciences sector is available in International Society for Pharmaceutical Engineering publications , such as the Good Automated Manufacturing Practice ( GAMP5 second edition ), which includes good practice in AI / ML implementation .
There are similarities with good practice for traditional coding but in the case of AI , good data governance , including data management , data quality and data selection practices , are essential right from the start .
The AI algorithms learn from an initial data set , a portion of which is also used to test and validate the application developed . These topics are covered by the new ITS GAMP 5 2nd edition training courses and implementation projects can be supported by our engineering and validation teams .
At ITS , our in-house experts are engaging with AI in a variety of ways and providing AI-related services to our clients . We have used AI to assess service level agreement call-out patterns and analysed the textual data to identify the root causes of increased weekend callouts , such as poor training and handover between shifts .
We have worked with a food sector manufacturer to investigate inconsistent product quality issues , exploring the potential for the application of AI to monitor the process using the JMP statistical analysis software . This included data sampling , an assessment of data quality , the improvement of data acquisition processes , targeted sampling using Design of Experiments and root cause analysis of process issues .
We are also partners with vision system supplier Cognex and are successfully exploring DL for the detection of defects and nonconformances that are not suited to conventional approaches in the life sciences and process sectors .
Our in-house expertise in IP21 systems and SQL makes us an ideal partner for data analytics . We believe AI implementations are key to improving productivity and look forward to working on new challenges with businesses in Teesside and across the UK .
The voice of business in the Tees region | 157