IBM and Pfizer’s AI tool can help manage the progression of Alzheimer’s better, with high-accuracy early detection
The mean time to the diagnosis of mild Alzheimer’s disease, as per the study, was 7.59 years.
Deployment of AI could mean a quantum leap for diagnostics—promising developments have already been reported for breast cancer, kidney disease and even Covid-19. But, neuropsychological pathologies have presented a different challenge altogether. A fortnight ago, however, a breakthrough was reported by IBM and Pfizer in the detection of the onset of Alzheimer’s. Using language data from the Framingham Heart Study of 1948, the technique detected speech impairment that is typical of Alzheimer’s. Observing 703 samples of 270 participants, the AI model predicted the probability of a person developing Alzheimer’s.
Data from the study highlights that linguistic models had a higher level of accuracy than non-linguistic determinants, demographic variables (age, gender, education, etc) and neuropsychological test results. The accuracy in the case of linguistic models was 73%. In contrast, non-linguistic determinants had an accuracy of 64%, whereas models which combined both linguistic and non-linguistic techniques had an accuracy of 72%. The prediction was easier in the case of females (84%) than males (63%). However, Alzheimer’s amongst college graduates was more challenging to determine than those who cleared just primary schooling.
The mean time to the diagnosis of mild Alzheimer’s disease, as per the study, was 7.59 years. Early detection can help manage the progression of the disease better. With AI sharpness depending on the quality and quantity of data made available, international collaborations can help map Alzheimer’s—as also other pathologies—more effectively. With high accuracy leading to better disease management, the outgo on, and the burden, of care for such diseases could come down meaningfully for nations.