{
  "abstract": "Background Neurofibromatosis type 1 (NF1) is a complex neurogenetic disorder marked by significant inter-patient heterogeneity in tumor growth, synaptic disruption, immune infiltration, and treatment response. Existing therapies, including MEK inhibitors, show variable efficacy due to incomplete understanding of the disease’s neuroimmune microenvironment and drug resistance mechanisms. There remains a critical need to stratify NF1 patients using bioinformatic and molecular features and to rationally design combination therapies based on individualized tumor and immune signatures. This study presents NeuroPharm-X+, an artificial intelligence (AI)-powered therapeutic simulation platform that integrates transcriptomic modeling, immune deconvolution, molecular docking, and logic-based pharmacodynamic simulation to identify and prioritize patient-specific dual-target therapeutic strategies.Methods Bulk RNA sequencing data from NF1-associated tumors were processed using DESeq2 for differential expression analysis and CIBERSORTx for immune cell deconvolution. Pathways involving IL6-STAT3, EGFR-PI3K-AKT, and GRM7-cAMP-CREB axes were prioritized based on transcriptional and immune heterogeneity. Molecular docking simulations were performed using AutoDock Vina with FDA-approved compounds from DrugBank and ZINC15, and pharmacophore filtering was conducted using RDKit and SwissADME. Systems biology models of pathway logic were constructed using PySB and BioNetGen to simulate downstream modulation of Syn1 phosphorylation, CREB activation, and mTOR signaling. A supervised machine learning model (XGBoost) was trained on expression and immune features to predict drug response probabilities across simulated patient cohorts. Performance was assessed using cross-validation and ROC analysis. Top drug combinations were ranked by synergy score, immune context, and pharmacologic viability.Results Analysis of NF1 transcriptomes revealed overexpression of IL6, EGFR, and CREB in distinct patient subsets, with immune deconvolution identifying macrophage-dominant and CD8-low phenotypes correlating with cytokine amplification. Docking simulations identified high-affinity FDA-approved inhibitors for IL6R and EGFR (e.g., tocilizumab and erlotinib), as well as PAM analogs targeting mGluR7. Logic simulation demonstrated that dual-target modulation of IL6 and EGFR, or mGluR7 and CREB, significantly reduced predicted activation of downstream effectors (figure 1). The machine learning classifier achieved an AUC of 0.93 in stratifying patients by predicted response to dual-target regimens (figure 2). Top-ranked therapeutic hypotheses were matched to real-world drug profiles for preclinical validation.Conclusions This project demonstrates a novel, translationally scalable approach to therapeutic design in NF1, combining AI-driven precision modeling with pharmacologic realism. This platform enables individualized, immune-informed combination strategies, supporting rational drug selection for future clinical trials in neurofibromatosis.Consent Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal.",
  "authors": [
    {
      "affiliations": [
        "Mind Matters Foundation, Flower Mound, TX, USA"
      ],
      "name": "Shivi Kumar"
    }
  ],
  "title": "1099 AI-guided therapeutic modeling of neurofibromatosis type 1 integrating immune profiling, cheminformatics, and logic-based dual-pathway suppression",
  "uid": "a85222bd-2162-50af-b82e-4d23c9371560"
}