A new system could help doctors predict what will happen to heart patients using technology originally designed for monitoring aircraft.
The software tool, developed by researchers at Lancaster and Manchester universities, provides a real-time analysis of the vital signs of hospital cardiovascular patients and compares them with a database of thousands of previous cases to predict the most likely outcome.
The researchers say the system, adapted from algorithms used to predict when aircraft will fail, could not only improve patient care but also help save the NHS money by preventing patients from being discharged from hospital too early.
‘Before releasing a patient, they would run the algorithm and say this is what will happen to the patient if he goes home,’ said project leader Prof Garik Markarian, an expert in aviation security at Lancaster University.
‘If we keep him an extra day or two in the hospital, how will we improve his chances of not coming back? This has financial implications because there will savings for the hospital by keeping a person one day extra in the hospital [rather than] having him readmitted.’
Making the software web based could also enable consulting doctors to make treatment recommendations even if they weren’t at the hospital, he added, provided adequate security could be implemented.
Markarian and his partners decided to build the system after discussing how a computerised solution could help analyse patients and noticing the similarity between the checklists used by pilots and heart surgeons before beginning a procedure.
Aircraft typically use a system that analyses data from up to 1,000 on-board sensors to determine when maintenance is needed. The researchers adapted this system’s algorithms to monitor heart rate, blood oxygen content and other vital signs, and make predictions in a similar way.
As reference, the team obtained a database of around 30,000 patient records from the Massachusetts Institute of Technology (MIT) in the US, removing irregularities and combining it with a smaller but more specialised database from Manchester.
The tool uses this database along with patient-specific information to provide a prediction of the patient’s condition for the next 24 hours. ‘The doctor will set the parameters and tune what is normal and what is high for that patient,’ said Markarian.
After six months of testing by academics at University Hospital of South Manchester led by Prof Charles McCollum, the researchers found the tool made accurate predictions in around 75 per cent of cases.
They say this could increase to 85 per cent by gathering more data and working with mathematicians to improve the algorithms and are applying for more funding to develop the system and ethical clearance to start large-scale trials with patients.
‘Potentially, it could be used for other things,’ said Markarian. ‘People from the urology department have shown a lot of interest and recently we’ve been talking to people from a trauma unit who would like to predict if an organ is going to fail, for example.’