A research coordinated by the CNR-Istc has used, for the first time, an algorithm of machine learning to analyze the outcome of neuropsychological, neurophysiological and genetic tests aimed at predating the onset of Alzheimer’s and Parkinsonon taking into account sex.
What role does sex has in the development of Neurodegenerative pathologies such as Alzheimer’s disease and Parkinson’s disease? To the question he tries to answer a study coordinated by the Institute of Sciences and Technologies of the Cognition of the National Research Council of Rome (CNR-ISCC), which for the first time used the tool of theArtificial intelligence (IA) To identify the most important factors for the early diagnosis, differentiating men and women.
In particular, they have been subjected to an algorithm of AI The outcome of a series of neuropsychological tests, neurophysiological and genetic data conducted on a mixed champion – composed of both healthy men and women and patients and, with the aim of Identify and differentiate on the basis of sex the main predictive factors associated with the onset of the two diseases.
The results of the research, the result of an interdisciplinary work that also involved the Milan 4 research area of the CNR, the Mondino Foundation, the University of Pavia, the Santa Lucia IRCCS Foundation, the Universities of Rome Sapienza and Tor Vergata and ai2life srl, a start-up developed within the CNR-Istc, are published in two distinct articles of the Journal of the Neurological Sciences. The two articles report the outcome of the tests conducted with the Machine Learning model to predict, respectively, the onset of Alzheimer’s disease and Parkinson’s disease.
The scientific manager of the research, Daniele Caligiore, research manager at the CNR-Istc and director of the Advanced School in Artificial Intelligence (As-ai), said:
The novelty of the study consists in having adopted an integrated approach in the analysis of the tests, consistently with the theory we developed at the CNR-Istc, according to which both pathologies -alzheimer and Parkinson-could be manifestations of a single disease, called neurodegenerative Elderly Syndrome (NES). In the analysis of the tests we started from analyzing the differences between healthy patients and sick patients, regardless of whether they were men or women: in fact, there are many studies that compare the outcome of the predictive tests on the basis of the genre, but do not consider that some characteristics can be relevant for both groups, regardless of the absolute values of test scores. Our research face this problem for the first time through an explanable machine learning algorithm, that is, capable of making the used decision -making process transparent, increasing reliability and promoting adoption in the medical field.
In the case of Alzheimer, the algorithm analyzed the outcome of simple neuropsychological tests aimed at estimating the probability of the onset of the pathology depending on the sex on the basis of P“predatory” Arameters such as memory, orientation, attention and language (Mmse); short -term verbal memory (AVTOT); And the long -term episodic memory (Lideltotal).
The machine learning system that we developed shows how MMSE is a more effective predictor than Alzheimer’s in women, while in men it is essential for long -term monitoring. Lideltotal is more predictive in women for the onset of the disease, while Avtot is more relevant in men. In addition, the level of education affects differently on Alzheimer’s risk, with women presenting greater risk.
The Machine Learning model developed for research on Parkinson’s Instead, he identified key characteristics – neuropsychological, genetic and body – which can be linked to the onset of the pathology. With regard to men it emerges that they are to be considered among the main predictors of the onset of Parkinsonon data that measure the muscle rigidity and dysfunctions of the autonomous nervous system; while for women the data on the urinary dysfunctions to predict the disease.
In addition, the Machine Learning model has identified the significant predictors of Parkinson’s also the age and family history of the champion, with a greater impact in men. Furthermore, they seem to be more relevant, always in the male sphere, the tests that measure the semantic verbal fluidity (SFT) and the data on the SNCA-TCA-GRES356181 genetic variant, linked to the alfa-sanuclein gene, a protein involved in the development of neurodegenerative diseases such as Parkinson’s.
The results of these research shows the importance of integrating specific diagnostic approaches for sex in clinical practice to improve the management of Alzheimer and Parkinson’s: the task of the research will be to refine more and more neuropsychological tests and predictive biomarkers, with particular attention to Sex so as to support personalized treatments. In addition, our study represents a concrete example of how IAs can effectively support medicine, combining the analysis of individual characteristics with a systemic vision: Machine Learning algorithms, in fact, can integrate and analyze specific patient – physiological data data , genetic or lifestyle linked – to predict the onset of the disease, monitor their progression and, at the same time, offer targeted and personalized treatments.
You may be interested in:
