Innovation: Screens to give patient advice

Click to follow
COMPUTER scientists are applying artificial intelligence techniques to provide patients with personal explanations of their medical records.

The aim is to develop information systems that automatically compose appropriate answers to patients' questions from a medical database and knowledge about individuals from patient records.

This will enable doctors to fulfil their legal requirement to give patients access to their medical records, without creating the extra burden of having to explain what the records mean. There is also evidence of a medical equivalent of the phenomenon of people preferring to use cash dispensers rather than bank tellers.

'Even people who are not familiar with computers find them less intimidating than asking their doctor the same questions,' says Alison Cawsey, who is developing the system at Glasgow University.

There should also be cost savings for the National Health Service - several studies have shown that patients given full details and access to information about their condition consult their doctors less often.

One carried out by Aberdeen University and reported in the British Medical Journal in February revealed 54 per cent fewer hospital admissions among asthma sufferers given specially prepared information about the condition, than among a control group of patients that did not receive the information.

Several computer-based health information systems have already been developed and their usefulness tested in large-scale public trials in Glasgow and elsewhere. But these systems are not based on patient records, and everyone who asks about a specific topic receives the same information.

Ms Cawsey and her colleague Kim Binstead are exploring how coherent and accessible texts should be composed, given knowledge of the patient.

For example, the system might need to 'explain what diabetes is' in the light of knowledge that the patient has just been diagnosed as having it.

'The system should be able to decide what is important for this patient and put the information together in a clear, understandable way,' Ms Cawsey says.

The work at Glasgow is focusing on patients with diabetes. The researchers are developing a computerised database about diabetes, and they will use anonymous records of real patients at a diabetes clinic in Nottingham to compose the individual explanations of medical terms and concepts.

For example, in noting symptoms of diabetes, the doctor would write down 'polydipsia', and the computer system would help the patient to interpret the record by explaining that the term means excessive thirst.

And while the doctor would only record the name of the medication chlorpropamide, the system would explain that chlorpropamide is a drug that promotes the production of insulin, and go on to explain that it is not recommended for patients with liver disease, kidney disease or endocrine disorders, and should not be taken with oral contraceptives, aspirin or antibiotics.

Patients will also be provided with a printout of their dialogue with the computer.

The researchers aim to develop a prototype system by the summer that can be tested on a small number of patients in an outpatient clinic.

(Photograph omitted)