Not a member yet? Register for full benefits!

Encroachment of the Computerised Doctor

After decades of future predictions that came to naught, it looks like the medical community is finally; very begrudgingly accepting that computerised physicians might actually be a desirable thing.

In late February 2008, researchers from the Wellcome Trust Centre for Neuroimaging at University College London ran a clinical trial for Alzheimers detection.

Expert systems, specifically trained for the task had a 96% diagnosis success rate analysing a clinical MRI scan for evidence of the onset of Alzheimers, compared to an 85% success rate for doctors using a combination of standard scans, blood tests and interviews.

Going back to the Wellcome trust Alzheimers detection AI, Professor Richard Frackowiak, one of the instigators of the study said the computers were better able to distinguish signs of Alzheimer's than humans, and proved cheaper, faster and more accurate than current methods.

"It's beginning to look like it will have to come into clinical practice," he said. "Machines are clearly able to do that sort of thing better."

Professor Richard Frackowiak said the computers were better able to distinguish signs of Alzheimer's than humans, and proved cheaper, faster and more accurate than current [human] methods.

"It's beginning to look like it will have to come into clinical practice," he said. "Machines are clearly able to do that sort of thing better."

In any area where analysis of complex data for minute changes that should not be there is required, it seems an analytical, tireless computer system consistently trumps human doctors.

A few years back, another system emerged, Ferriscan. Ferriscan is a computer imaging system to detect iron ferrite build-up in the body. It uses this to diagnose the state of a patient's liver.

FerriScan uses existing MRI (magnetic-resonance imaging) machines to produce its data, then frequency shifts the results, and colour-maps. It has been used clinically in the US, UK, and Australia. It, like almost all medical imaging systems, currently relies on human doctors to make judgement calls on the results. However, if additional clinical case studies come back as with the example above, that the AI system can identify results consistently, at a superior rate to humans - and there is no reason to believe they would not - then the amount of person hours necessary with multiple consultations to identify a condition, drops off sharply.

Patient interviews are nice, but they could be conducted long-distance, and simply to check data, if the patient so desires. Working alongside humans, such diagnostic systems, if put in place would dramatically slash the cost of provision of medical care.

As empirical results begin to slowly trickle back, and it becomes more and more obvious that the AI based approach is better than the human based approach, in terms of both patient care and expenditure, traditionally stodgy health services are being forced to re-examine their positions that a computer has no place in decision making about patient healthcare.

Leading on from this, is the continual, steady rate of adoption of the twin standards of PACS and DICOM. PACS is the British medical imaging standard, or Picture Archiving and Communications System. It takes X-rays, radiotherapy, CT, MRI, nuclear medicine, angiography, cardiology, fluoroscopy, ultrasound, dental and symptomatic mammography, combining all into one single image standard.

DICOM does much the same. It is the American standard; Imaging and Communications in Medicine.

Between the two they standardise medical data into a common format. This makes it much, much easier for machine vision systems to then study the data files, combine them, and analyse them, without human intervention.

Computerised doctors will not completely replace human physicians of course. Most patients prefer that human touch when they go to the doctor, or, when the doctor comes to them, in the form of telehealth.

Telehealth basically being remote healthcare, with the doctor's image, and patient's image + vital signs being transmitted to one another, without actual physical travel. This saves time and costs for the doctor, without compromising patient care - the patient still has a secure way to express feelings and concerns. Telehealth is actually extremely beneficial in other cases, for instance where the patient has a condition that emits an unpleasant aroma beyond their control. The doctor is aware of it, via remote sensors, but their own nose is not subject to it, making the environment more pleasant and less embarrassing for both.

How it might work:

The patient goes to a doctor's surgery. Rather than seeing a doctor directly, they slip into one of many cubicles. These cubicles are shared by many different surgeries, pooling the doctors, with specialists on hand. A doctor talks to the patient, and maybe recommends they see a specialist. The doctor contacts a hospital, and secures a link to a specialist to discuss the options three-ways during the same appointment.

The specialist recommends the patient go to a hospital for testing.

At the hospital later on, they run a complete set of scans, and send the patient home. Before the patient has crossed the car part, the images are on their way into various AI systems, where they are examined, manipulated, and overlaid. The AIs have reached a decision as to whether or not the patient is suffering from the issues each is trained to look for, maybe before the patient's car is started, and the results transferred via Internet to their patient notes before the car has left the grounds.

Next appointment with the doctor a few days later, the test results are ready, and clear. Proposed treatment regimens for the patient's personal history are also prepared, which the doctor hasn't seen before this time. Raw data is there if there are any queries, but that will take time to sift through.

The doctor is not really needed, save for the psychological comfort of talking to a human. In time, that too, may be unnecessary for the more mundane illnesses, freeing up general practitioner time, for more difficult to diagnose complaints.

Staff Comments


Untitled Document .