Artificial Intelligence in medicine
Machines do make mistakes. Until now we call them Artefacts, but they are limited to findings in an ECG or a MRI scan. But how do we find out an Artefact in a clinical diagnosis or a complex decision of Artificial Intelligence ?
In 2015, a research group at Mount Sinai Hospital in New York applied Artificial Intelligence to the hospital’s vast database of patient records.
The program ‘Deep Patient’ was created by feeding it hospital data of about 700,000 individuals: hundreds of variables of these patients, their test results, case records, doctors’ assumptions, course of disease and final results.
The AI doctor proved incredibly good at predicting disease.
Joel Dudley, who leads the Mount Sinai team says, “Deep Patient’ is a bit puzzling. It appears to anticipate the onset of psychiatric disorders like schizophrenia surprisingly well. It had discovered patterns hidden in patient records missed by treating doctors.”
Of course he doesn’t know how is it possible ? The new tool offers no clue as to how it does this. In a very large neural network like ‘Deep Patient’ there are thousands of units per layer and maybe hundreds of layers. They trawl gigabytes of data and millions of calculations before reaching a pattern recognition and complex decision-making.
If the AI program makes a mistake it will be very difficult to figure it out.
Machines do make mistakes. Until now we call them Artefacts, but they are limited to findings in an ECG or a MRI scan. But how do we point that Artefact out in a clinical diagnosis or decision of AI ?
Dr Ramakanta's Ideas 