| "A computer system that can rival some doctors' ability to diagnose early-stage Alzheimer's is being trialled in the US. It analyses a person's brain waves as they tackle a number of simple sound-based tests." "An estimated 24 million people worldwide suffer the dementia caused by Alzheimer's disease. Early diagnosis is difficult and remains one of the greatest challenges in both patient care and the search for new treatments." "The only way to make a definite diagnosis is to find the telltale plaques of protein in a person's brain after death. Prior to this, memory and cognition tests are used to diagnose sufferers. But not everyone gets access to the right expertise and a US study published in 1999 found that community health centres only spot around 75% of Alzheimer's cases, while specialist centres, including large hospitals, are 85% to 90% accurate." "Now a team of US researchers have developed a computer system that could help close this gap. Developed by computer engineer Robi Polikar of Rowan University, in New Jersey, neurologist Christopher Clark from the Alzheimer's Disease Center in Pennsylvania and psychologist John Kounios at Drexel University in Philadelphia, the machine shows better accuracy than the average community centre." "It uses an electrode cap to monitor a person's brain waves as they take a sound-based test. The person is played a series of low tones, with a much higher "oddball" tone occurring intermittently. When the oddball tone occurs, they must press a button, and their brain's response is also recorded." "In healthy people, the oddball tone causes a spike in brain activity about 300 milliseconds after it occurs, known as the P300 response. In the new system, software looks for alterations in the strength and timing of this spike across a range of brainwave frequencies. In general, the peak is weaker and later in patients diagnosed with early-stage Alzheimer's." "In trials, the researchers recruited 28 people judged to be suffering from early stages of the disease and 24 who were not � 52 in all � and put them all through the test." "The system then analysed the entire data set 52 times. Each time, it used 51 of the patients' information to learn the differences between the two groups from scratch
... read the whole article |