AI Identifies Gender-Specific Factors in Alzheimer's and Parkinson's Prediction

A study coordinated by Cnr-Istc [National Research Council - Institute of Cognitive Sciences and Technologies] utilized machine learning to analyze neuropsychological, neurophysiological, and genetic tests, considering sex, to predict Alzheimer's and Parkinson's disease onset. The research aims to establish a foundation for gender-specific diagnostic approaches in clinical practice. Findings are published in the Journal of the Neurological Sciences. The study, coordinated by the Cnr-Istc [National Research Council - Institute of Cognitive Sciences and Technologies] in Rome, employed Artificial Intelligence (AI) to identify key factors for early diagnosis in men and women. A mixed sample of healthy and affected individuals underwent neuropsychological tests, neurophysiological data collection, and genetic analysis. The AI algorithm then identified and differentiated the primary predictive factors associated with the onset of both diseases based on sex. The research, involving the Cnr's [National Research Council] Area di Ricerca Milano 4, Fondazione Mondino, Università di Pavia, Fondazione Santa Lucia IRCCS, Università di Roma Sapienza and Tor Vergata, and AI2Life s.r.l., published results in two Journal of the Neurological Sciences articles. These articles detail the machine learning model's performance in predicting Alzheimer's and Parkinson's. Daniele Caligiore, Research Director at Cnr-Istc [National Research Council - Institute of Cognitive Sciences and Technologies] and Director of the Advanced School in Artificial Intelligence (AS-AI), stated: "The study's novelty lies in its integrated approach, consistent with the theory that both Alzheimer's and Parkinson's might be manifestations of a single disease, termed Neurodegenerative Elderly Syndrome (NES)." The AI algorithm analyzed differences between healthy and ill patients, regardless of sex, using explainable machine learning to increase reliability and promote medical adoption. For Alzheimer's, the algorithm analyzed neuropsychological tests estimating disease probability based on sex, using parameters like memory, orientation, attention, and language (MMSE [Mini-Mental State Examination]); short-term verbal memory (AVTOT [Auditory-Verbal Test]); and long-term episodic memory (LDELTOTAL [Long-Delay Free Recall Total]). The researcher noted, "The machine learning system shows MMSE is a more effective predictor in women, while in men it's essential for long-term monitoring. LDELTOTAL is more predictive in women, while AVTOT is more relevant in men. Education level also impacts Alzheimer's risk differently, with women facing a greater risk." The Parkinson's research model identified key neuropsycological, genetic and physical characteristics linked to disease onset. Muscle rigidity and autonomic nervous system dysfunctions are major predictors for men, while urinary dysfunctions are more relevant for women. Age and family history were also significant predictors, with a greater impact in men. Verbal semantic fluency (SFT [Semantic Fluency Test]) and the genetic variant SNCA-rs356181, linked to the alpha-synuclein gene, were more relevant in men. Caligiore stated: "These results highlight the importance of integrating gender-specific diagnostic approaches to improve Alzheimer's and Parkinson's management. Future research will refine neuropsychological tests and predictive biomarkers, focusing on sex to support personalized treatments." He added, "Our study demonstrates how AI can effectively support medicine by combining individual characteristics with a systemic view, integrating patient-specific data to predict disease onset, monitor progression, and offer targeted treatments."