Artificial Intelligence-Assisted Comparative Analysis of the Overlapping Molecular Pathophysiology of Alzheimer's Disease, Amyotrophic Lateral Sclerosis, and Frontotemporal Dementia

Abstract

The overlapping molecular pathophysiology of Alzheimer's Disease (AD), Amyotrophic Lateral Sclerosis (ALS), and Frontotemporal Dementia (FTD) was analyzed using relationships from a knowledge graph of 33+ million biomedical journal articles. The unsupervised learning rank aggregation algorithm from SemNet 2.0 compared the most important amino acid, peptide, and protein (AAPP) nodes connected to AD, ALS, or FTD. FTD shared 99.9% of its nodes with ALS and AD; AD shared 64.2% of its nodes with FTD and ALS; and ALS shared 68.3% of its nodes with AD and FTD. The results were validated and mapped to functional biological processes using supervised human supervision and an external large language model. The overall percentages of mapped intersecting biological processes were as follows: inflammation and immune response, 19%; synapse and neurotransmission, 19%; cell cycle, 15%; protein aggregation, 12%; membrane regulation, 11%; stress response and regulation, 9%; and gene regulation, 4%. Once normalized for node count, biological mappings for cell cycle regulation and stress response were more prominent in the intersection of AD and FTD. Protein aggregation, gene regulation, and energetics were more prominent in the intersection of ALS and FTD. Synapse and neurotransmission, membrane regulation, and inflammation and immune response were greater at the intersection of AD and ALS. Given the extensive molecular pathophysiology overlap, small differences in regulation, genetic, or environmental factors likely shape the underlying expressed disease phenotype. The results help prioritize testable hypotheses for future clinical or experimental research.

Keywords: Alzheimer’s disease (AD); amyotrophic lateral sclerosis (ALS); frontotemporal dementia (FTD); knowledge graph; literature-based discovery; machine learning; network regulation; neuropathology; neurophysiology; pathology dynamics.

Figure 1

A Venn diagram of three diseases (AD, ALS, and FTD) illustrating intersections and unions for AAPP (amino acid, peptide, proteins) node type. FTD is represented by the light yellow circles; AD is represented by the aqua blue circles; ALS is represented by the light red circles. Intersections are shown in percentages for each disease. (A) FTD; (B) AD; (C) ALS.

Figure 2

CompositeView of top 1% AAPP (amino acid peptide protein) nodes from AD, ALS, and FTD. CompositeView is a visualization tool for SemNet 2.0 that allows for the aggregation and visualization of multiple simulation outputs into a compressed form [24]. The source nodes are colored in green, while the target nodes are colored in blue with names next to them. There are large amounts of source nodes shared only between AD and ALS; meanwhile, the AD and FTD and ALS and FTD disease pairs (circled in orange) fewer nodes. The majority of nodes are shared by all three diseases. Nodes exclusive to only one disease are labeled outside the apex: AD has many exclusive nodes; ALS has a few exclusive nodes; and FTD had no exclusive nodes.

Figure 3

Sunburst diagram of high-ranking AAPP nodes for each disease intersection mapped to their biological processes. The width of each segment represents the number of source nodes. The inner ring shows the relative counts of high-ranking intersecting nodes belonging to each biological process. Clockwise from the top of the inner circle: synapse and neurotransmission (dark green); inflammation and immune response (navy); cell cycle regulation (dark purple); protein aggregation (green); membrane regulation (turquoise); energetics and metabolism (dark orange); stress response regulation (brown); gene regulation and expression (obsidian). The outer ring shows the relative counts of the mapped biological processes for each disease intersection: AD ∩ ALS, AD ∩ FTD, ALS ∩ FTD.

Figure 4

Bar chart quantitatively illustrating the biological mapping of the top 1% of high-ranking AAPP nodes based on all disease intersections: AD ∩ ALS, AD ∩ FTD, and ALS ∩ FTD.

Figure 5

Bar chart illustrating the normalized biological mapping of the top 1% of high-ranking AAPP nodes adjusted for differences in node count between intersecting diseases: AD ∩ ALS, AD ∩ FTD, and ALS ∩ FTD.

Figure 6

Z-scores to examine standardized differences in the biological process mappings of disease intersections compared to the overall average for each biological process: AD ∩ ALS, ALS ∩ FTD, and AD ∩ FTD.

Figure 7

Overall framework for artificial intelligence-based comparative analysis of the molecular pathophysiology overlap of AD, ALS, and FTD. Over 33 million journal articles from PubMed were used to construct the knowlege graph used by SemNet 2.0. The open-source SemNet 2.0 software [21] and its post-visualization software, CompositeView [24], were used to determine which Unified Medical Lanaguge System (UMLS) amino acid, peptide, and protein (AAPP) nodes are most important to AD, ALS, FTD, and, namely, their intersections. Next, the top 1% of nodes were mapped to eight biological processes using cross-domain text mining natural language processing and a large language model. Finally, three human evaluators provided a check on the top 1% of nodes via full text article review and on the corresponding biological mappings.