Professor Jinhua Sheng and his team from Hangzhou University of Electronic Science and Technology have recently made breakthroughs in understanding the onset and progression mechanisms of Alzheimer's disease (AD), with their research published in both the journals Cerebral Cortex and Computers in Biology and Medicine.
Currently, approximately 50 million people worldwide suffer from Alzheimer's disease (AD), and it is projected that the number of AD patients will exceed 150 million by 2050. However, effective preventive and therapeutic measures for AD are still lacking, making early diagnosis and prevention of AD a pressing scientific and societal issue.
In their study published in Cerebral Cortex, Sheng and his collaborators found a close correlation between neuroinflammation-mediated neuronal dysfunction and the activation of Macrophage Stimulating Protein 1 (MST1). They proposed the use of the correlation between fluorodeoxyglucose positron emission tomography (FDG PET) and amplitude of low-frequency fluctuations (ALFF) as imaging markers for the overactivation of microglia and neuronal damage associated with neuroinflammation. Experimental results showed that among 121 patients with cognitive impairment, those carrying the risk allele of MST1 rs3197999 exhibited significantly reduced coupling of glucose and oxygen metabolism in the prefrontal cortex region. Additionally, the risk allele of rs3197999 was highly correlated with an increased rate of clinical dementia score, mediated by the coupling of glucose and oxygen metabolism. Sheng indicated that from a clinical perspective, MST1 rs3197999 may serve as a target for drug development, offering a new avenue for clinical treatment of AD or other neurodegenerative diseases.
In their study published in Computers in Biology and Medicine, Sheng et al. addressed the limitations of single-modality neuroimaging data in AD diagnosis by proposing a multi-modal machine learning framework that integrates complementary biomarker data. This model combines magnetic resonance imaging, positron emission tomography, and cerebrospinal fluid detection results to enhance AD characterization. The researchers proposed a hybrid algorithm that combines improved group intelligence optimization with extreme learning machines to simultaneously perform feature selection and classification. The study demonstrates the advantages of utilizing complementary multi-modal data to improve AD diagnosis through advanced feature learning techniques.
Related Paper Information:
