AI-Driven Repurposing of FDA-Approved Compounds for Precision Neuroscience Therapeutics

Authors: Yassir Boulaamane, Mena Abdelsayed

1. Overview

(1) Scientific Rationale and Importance

Neurological disorders remain among the highest global unmet medical needs, with major therapeutic gaps in:

Drug-resistant epilepsy

Amyotrophic lateral sclerosis (ALS)

Frontotemporal dementia (FTD)

Alzheimer’s disease (early-stage disease-modifying therapy)

Treatment-resistant depression

Post-traumatic stress disorder (PTSD)

Neuropathic pain

Despite extensive target discovery, de novo drug development in neuroscience has a >90% failure rate. Repurposing FDA-approved compounds offers:

• Known safety profiles

• Established pharmacokinetics

• Reduced translational risk

• Shortened time to Phase II trials

Building on prior work in sodium channel pharmacology, precision electrophysiology, AI-driven polypharmacology, and iPSC disease modeling , we propose a scalable repurposing pipeline combining:

• Structure-based virtual screening

• AI-driven polypharmacology prediction

• Patient-derived iPSC neuronal models

• Electrophysiological phenotyping

• Mechanism-driven prioritization

(2) Funding / Facilities / Ethics

• Institutional lab space and stem cell facilities available at Lankenau Institute for Medical Research.

• iPSC differentiation infrastructure and electrophysiology platforms established (MEA, patch clamp).

• Computational screening resources via GPU-enabled infrastructure (NVIDIA Inception affiliation) .

• IRB approval will be obtained for any patient-derived cell lines.

• All compounds selected will be FDA-approved and commercially available.

(3) Estimated Timeline

Phase Timeline

In silico screening & prioritization Months 0–6

In vitro validation (neuronal iPSCs) Months 6–18

Mechanistic electrophysiology & omics Months 12–24

Lead nomination for IND-enabling strategy Months 18–24

(4) Data Sharing Commitment

• Preclinical datasets deposited in GEO / ArrayExpress.

• Chemical-target interaction matrices shared via open repositories.

• Negative results included to improve field reproducibility.

• Publication in open-access venues where feasible.

2. Introduction

Theoretical Background

Neuroscience drug development suffers from:

• Target reductionism

• Poor translatability

• Inadequate human-relevant models

Recent literature supports:

• Sodium channel modulation in epilepsy and ALS

• mTOR/autophagy modulation in neurodegeneration

• Neuroinflammation as a convergent pathway

• Mitochondrial stabilization as disease-modifying strategy

• Polypharmacology outperforming single-target drugs

Your prior expertise in:

• Voltage-gated sodium channel pharmacology

• Precision electrophysiology

• AI-driven polypharmacology

positions this project uniquely.

High-Unmet Need Indications (Ranked by Strategic Opportunity)

1. Drug-resistant epilepsy

2. ALS

3. Frontotemporal dementia

4. Early Alzheimer’s disease

5. Neuropathic pain

6. Treatment-resistant depression

Candidate FDA Drug Classes for Repurposing

1. Sodium Channel Modulators

• Ranolazine

• Lacosamide

• Mexiletine

• Flecainide (low-dose CNS exploration)

Rationale:

• Nav1.2 / Nav1.6 hyperexcitability in epilepsy

• Persistent INa implicated in ALS motor neuron stress

2. mTOR / Metabolic Modulators

• Metformin

• Rapamycin analogs

Rationale:

• Autophagy enhancement in FTD and ALS

• AMPK activation neuroprotection

3. Anti-Inflammatory CNS-Penetrant Agents

• Minocycline

• Ibudilast

• Dimethyl fumarate

4. Mitochondrial Stabilizers

• Nicotinamide riboside

• Pioglitazone

5. Polypharmacologic Oncology Agents (Low-dose)

• Dasatinib (senolytic)

• Imatinib (neuroinflammatory modulation)

Research Questions

1. Can AI-driven multi-target scoring identify FDA-approved compounds that outperform single-target strategies?

2. Do selected compounds normalize hyperexcitability in patient-derived neurons?

3. Are polypharmacologic signatures predictive of functional rescue?

Hypotheses

H1: FDA-approved compounds with multi-target ion channel + inflammatory signatures will outperform single-target agents.

H2: Compounds reducing persistent sodium current will reduce hyperexcitability in iPSC-derived neurons.

H3: Compounds activating AMPK/autophagy will reduce TDP-43 aggregation in ALS models.

Impact Statement:

Even null findings will refine computational filters and improve translational neuroscience screening paradigms.

3. Methods

Materials and Procedures

Step 1: AI/Computational Screening

• Docking against:

o Nav1.2

o Nav1.6

o mTOR

o TDP-43 interaction interfaces

o NF-κB pathway proteins

• Polypharmacology network scoring

Step 2: Functional Validation

• Multi-electrode array recordings

• Patch clamp electrophysiology

• Calcium imaging

Step 3: Mechanistic Assays

• Western blot (TDP-43 aggregation)

• Autophagy markers (LC3-II)

• RNA-seq for inflammatory signatures

Planned Analyses

Preprocessing:

• Spike detection normalization

• Batch correction

• Outlier filtering (ROUT method)

Primary statistical tests:

• Mixed-effects ANOVA

• Multiple comparison correction (FDR)

• Cohen’s d effect size reporting

IF-THEN Decisions:

IF compound reduces firing rate >30% without toxicity

→ advance to mechanistic profiling

IF cytotoxicity >20%

→ exclude from further testing

IF electrophysiology improves but inflammation unchanged

→ classify as excitability-specific candidate

Ethics and Data Management

• IRB approval prior to patient-line experiments

• De-identified cell line coding

• Secure encrypted data storage

• 10-year archival plan

4. Pilot Data (Optional)

Preliminary in silico screening (AML repurposing project with Boulamane) demonstrates feasibility of:

• Docking FDA library

• Ranking by binding affinity

• Performing MD simulations

These computational pipelines are directly transferable to neuroscience targets.

5. Budget

Provide an itemized list of anticipated costs (materials, personnel, etc.). These are not binding and are able to be adjusted based on the results of the pilot data.

6. References

Abdelsayed, M., & Boulaamane, Y. (2025). Drug Repurposing for AML: Structure-Based Virtual Screening and Molecular Simulations of FDA-Approved Compounds with Polypharmacological Potential. Biomedicines, 13(11), 2605.

AI-Driven Repurposing of FDA-Approved Compounds for Precision Neuroscience Therapeutics

AI-Driven Repurposing of FDA-Approved Compounds for Precision Neuroscience Therapeutics

Funding Progress

Related Funding

Request for Proposals - Neuroscience

Nonprofit Recipient

Topics

DOI