Close
The Problem
Antibiotics save lives, but the more they are used, the less effective they become. Infections learn how to resist them. More than 50,000 people in the UK have infections that resist antibiotics, and this number will increase if we do nothing.
In a typical hospital around 150 patients are taking antibiotics on any one day. Research suggests that 30-50% of individual prescriptions could be saved if they were reviewed by a specialist in antibiotics looking at things with the danger of over-use in mind. However, there aren’t currently enough of these specialists - microbiologists and antibiotic pharmacists - to review every antibiotic prescription and provide feedback to every doctor.
The Solution
It’s simple - use fewer antibiotics. But antibiotics seem so helpful that this hasn’t worked, the use of antibiotics in UK hospitals is still growing. Led by Dr Laura Shallcross, UCL Institute of Health Informatics, the research team aims to develop and test a system called SamurAI. This involves computers working in a way that copies human thinking (‘artificial intelligence’) and it takes advantage of the introduction of electronic patient records into UK hospitals. The system will combine historical data for patients prescribed antibiotics with the findings of specialists in infection as they review prescriptions. Under specialist medical supervision, the computer system will ‘learn’ what works best and when to start, stop or change the use of antibiotics. Once up and running the system should be able to identify patients who should be reviewed by specialists and guide antibiotic treatment for individual patients. The information produced for the specialists will be quite technical but we want also to try to provide more easily understood feedback, guidance and suggestions for individual doctors in the hospitals on the way they prescribe and how they explain their prescribing to patients. The research will start at one hospital in London, but the aim is to roll it out to 3 different hospitals within 4 years.
Evaluation and Safeguards
The project will evaluate the software’s effectiveness by comparing its predictions to what an expert would do. If it shows promise, the team will progress to testing the system’s ability to guide doctors to better prescribing, through carefully designed trials. It is estimated that the software can reduce antibiotic use by 10-20%, benefitting patients by reducing their time in hospital and the risk of side-effects from antibiotic treatment. Reduced use will also prolong the lifespan of our existing antibiotics.
The project will involve patients, the public and healthcare workers at all stages of the project, to understand their concerns about using patient data and agree how to protect it. Research findings will be shared with patients through a series of workshops.