Rx License-Rx

19-8759-tpNCS

The g3mclass Software for Multiclass Disease Diagnosis and Reporting.

Diagnostic software to stratify patients on biomarkers and therapeutic targets. Problem: Targeted therapy has revolutionized the treatment of cancer. The efficacy of such therapy is highly dependent on target levels that may increase, decrease, or remain unchanged in the tumor compared to healthy tissue. In clinical practice, the target’s status may be tested through methods such as immunohistochemistry (IHC), which relies on a trained professional to group samples using an arbitrary cutoff. Such a semi-quantitative approach may lead to over-/under- estimation of the target expression and ineffective health interventions. Molecular assays can measure multiple biomarkers but lag in multiclass diagnostic capabilities. Solution: The inventor developed an original software that uses tissue biomarkers and rigorous statistical analysis to classify patients by target expression with speed and precision. Inventors: Marina Guvakova.

Intelligence Memo

Owner: University of Pennsylvania

Core category: Therapeutics

Therapeutic area: Oncology

Indication: Diabetes

Modality: Small Molecule

Focus tags: Oncology, Immunology, Neurology, Cardiometabolic

Technology tags: Small Molecule, Biologic, Diagnostic / Biomarker, AI / ML, Cell/Gene Therapy

Mechanism:

Development stage: Preclinical

Patent status: Copyright

Availability: Available for license

Plain-English Licensing Breakdown

This is a license opportunity for a therapeutic asset or drug-enabling technology in Oncology. In plain English, the buyer would be licensing science that could become a treatment program, usually after more validation. The current package appears to be preclinical and is associated with University of Pennsylvania. The practical first use case is Diabetes. Public description: Diagnostic software to stratify patients on biomarkers and therapeutic targets. Problem: Targeted therapy has revolutionized the treatment of cancer. The efficacy of such therapy is highly dependent on target levels that may increase.

What is exciting

Already past pure discovery: Preclinical validation gives a buyer something concrete to reproduce, optimize, or package into an IND-enabling plan.

Oncology remains highly partnerable: Pharma buyers still pay attention when an asset can be tied to biomarkers, combinations, resistance biology, or a defined tumor segment.

High upside if the mechanism is measurable: Neurology is hard, but biomarkers, retinal surrogates, genetics, or target-engagement readouts can turn a vague CNS story into a fundable experiment.

Can sell into pharma before reimbursement: A biomarker or AI tool can create value as trial enrichment, patient stratification, or translational support before becoming a regulated diagnostic.

Negatives / diligence concerns

Translation still unproven: Animal or lab data may not predict human performance; tox, PK/PD, CMC, and indication selection still need diligence.

Validation can be harder than the demo: Models and biomarkers need locked datasets, external validation, clinical utility, data rights, and a regulatory/reimbursement plan.

Manufacturing can dominate the budget: Potency assays, vector or cell process reproducibility, release testing, and COGS can become bigger risks than the biology.

CNS translation is unforgiving: Brain exposure, target engagement, endpoint sensitivity, and placebo/noise risk can make development expensive without a biomarker-first plan.

Risk Flags

  • Human validation and clinical path require diligence.
  • Patent scope and remaining exclusivity need review with counsel.
  • Inventor readiness and licensing terms are not yet verified.

Strategic Pharma Attractiveness

Large pharma would care if this becomes more than an interesting university-originated technology: it needs a crisp Oncology wedge, a measurable value inflection, and a diligence package that makes the first deal feel like an option on upside rather than a blind research bet.

Most logical pharma targets BMS / 2seventy — Cell therapy portfolio logic; needs differentiated antigen strategy. Gilead / Kite — Manufacturing and oncology BD infrastructure already exists. Regeneron — Deep oncology biologics and T-cell engager adjacency.

Development Strategy to Increase PoS

First indication: Diabetes

Study design: Retrospective locked-dataset validation followed by one prospective pharma enrichment pilot.

Key experiments Validate the AI-optimized pivot: Convert CNS risk into a measurable metabolic-rescue or peripheral biomarker strategy Run independent replication of the core claim with pre-specified success criteria Generate a partner-facing risk register that separates solved, testable, and unresolved risks

Final Recommendation

Proceed: Worth a short exclusive option if diligence confirms IP scope and inventor data quality. The most investable version is: Convert CNS risk into a measurable metabolic-rescue or peripheral biomarker strategy

Best next experiment: Run the smallest independent study that validates: Pair the asset with a brain-bioavailable precursor, nasal/local delivery, or exosome/nanoparticle carrier and gate spend on biomarker movement.

Best licensing timing: Begin BD conversations after the next validation package; pursue a license, option, or asset sale once the first value inflection is visible.