Tag: machine learning

Welcome to William Bolton

Profile photo

CAMO welcomes Will Bolton as he starts his AI4Health CDT-funded PhD studentship, supervised by Professor Alison Holmes and Dr. Pantelis Georgiou.

As part of the programme, Will’s PhD will combine artificial intelligence and machine learning to support healthcare diagnosis, monitoring and improvements in efficiency of care delivery.

The project aims to develop intelligent, personalised clinical decision support systems in order to optimise antibiotic therapy in obesity patients.

When an individual has two or more chronic healthcare conditions, this is known as multi-morbidity and this presents a significant challenge in healthcare. Obese patients are at higher risk of acquiring infections and being prescribed complex antibiotic treatments. They often fail to be dosed appropriately due to a lack of evidence to support decision making. This leads to a major health equality in management and outcomes of infections within obese cohorts, including risk of antimicrobial resistance. Will’s project will develop a novel decision support system, linking artificial intelligence with data from electronic health records to improve the diagnosis and management of infections in obese patients.

Tackling Antimicrobial Resistance with Artificial Intelligence

Researchers from the Centre for Antimicrobial Optimisation (CAMO) and the Centre for Bio-Inspired Technology at Imperial College London (ICL) have developed a new data-driven method to increase multiplexing capabilities of widely used PCR instrumentation.

In two studies published last month, the team at ICL, demonstrated the method using single-molecule real-time PCR. This increases the throughput of molecular diagnostic platforms and reduces the cost of tests, without any changes to instrument hardware, by virtue of smarter data analytics.

Artificial Intelligence DNA

“There is plenty of room to maximise the value of existing data using sophisticated machine learning methods.”

-Dr Jesus Rodriguez-Manzano

CAMO Chief Scientist

In the first study, the team explored ways to enhance multiplexing capabilities by training machine learning models using the kinetic information in DNA/RNA amplification curves. As a proof-of-concept study, this was shown using a 3-plex assay targeting common carbapenemase genes (KPC, NDM and VIM). In the second study, the group incorporated thermodynamic information via melting curves, to enhance the method to high-level multiplexing applications, such as detecting nine variants of mobilised colistin resistance (mcr-1 to mcr-9).

For more information, read media coverage of the new methods from GenomeWeb.

 

written by:

Jesus Rodriguez-Manzano