Artificial intelligence and Alzheimer's disease
PARIS HAMLATZIS, SMOLOF-ALEXANDROS HAMLATZIS

According to different studies, 33% of all dementia cases are due to some reversible disorder (hypothyroidism, B12 Vitamin deficiency, etc) while the remaining 67% to some neurodegenerative disease. A 90% of the victims of this second group is suffering from Alzheimer΄s disease. Bearing in mind that the different therapeutic interventions are likely to decrease the evolution of the disease only in mild and moderate cases, early diagnosis acquires a great significance. In this scope the contribution of artificial intelligence in the form of artificial neural networks seems very important.

Based on a number of remarkable studies, it seems that artificial neural networks offer a great potential in improving the diagnostic accuracy of those examinations that present complex patterns of data, such as positron emission tomographies (PET), single photon emission tomographies (SPECT), etc, which are difficult to discern even for an expert observer. Their ability to detect higher order non linear relations among the data they process (which are used as indices for the differentiation of pathological cases), renders them powerful classification tools with performances equal to those of expert scientists. In many cases these systems outrank several statistical methods available.

Key words: Artificial intelligence, Alzheimer’s disease.