BNP BioHybrid Neural Processor

BioHybrid Neural Processor · UIPS, Panjab University

Test CNS drugs on living human neural networks.

BNP grows human iPSC-derived neurons on a microelectrode array, records their electrical activity, and uses machine learning to read how a compound changes the network. The aim is an earlier, human-relevant signal for efficacy and seizure risk.

Run the virtual assay Project dashboard ↗
Substrate
Human iPSC cortical neurons with astrocytes
Interface
60-channel MEA, scaling to high-density arrays
Readout
Spikes, bursts, synchrony, decoded by ML
Simulated activity · 60 electrodes · 200 µm pitch · 30 µm Ø

The problem

Most CNS drugs fail in the clinic. The models used to choose them are part of the reason.

Neurology and psychiatry have some of the lowest success rates in drug development. Candidates that look promising in rodents and simple cell cultures often fail on efficacy or safety once they reach patients.

5.9%

of neurology drugs entering Phase I went on to approval, 2011–2020

BIO, Informa & QLS, 20215
99.6%

of Alzheimer’s drug candidates tested in trials failed, 2002–2012

Cummings et al., 20146
~98%

of small-molecule drugs do not cross the blood–brain barrier

Pardridge, 20057
3.4 B

people were living with a neurological condition in 2021

GBD 2021, Lancet Neurology8
$2.6 B

capitalised R&D cost per approved new drug, all therapy areas

DiMasi et al., 20169
192 M

animals estimated to have been used in research worldwide in 2015

Taylor & Alvarez, 201910

Root cause

Rodent neurons, tumour-derived cell lines and computer models each miss part of human neural biology: receptor subunits, ion-channel expression and the way real networks fire together. A compound can look safe or effective before the clinic and still fail in people. A human cell model with a functional electrical readout closes part of that gap at the stage when dropping a compound is still cheap.

Platform architecture

Three layers, from living cells to a drug profile.

Signal moves down the stack: neurons fire, electrodes record the voltage, and models turn the recordings into features and predictions. Select a layer to see what it does and how it is specified.

Mechanistic pathway

From compound to pharmacodynamic output, in six steps.

Each experiment follows the same sequence. Step through it, or let it play.

See steps 4–6 in the virtual assay

Virtual assay

Dose a network and watch the decoder read it.

Choose a reference compound and a concentration, and see how firing, bursting and synchrony change. Turn on blind mode to check whether the decoder can name the mechanism without being told the compound.

Well B3 · hiPSC cortical network · DIV 42 · 37 °C
Compound

Spike raster · 60 electrodes · last 10 s spikenetwork burstpopulation rate
Electrode map8 × 8

Bursts start at one point and spread across the array in tens of milliseconds. Dimmed electrodes have stopped recording activity.

Decoder readoutLive

Most likely mechanism

—

    Flags are hazard signals for follow-up, not safety conclusions. Loss of activity can also mean cell death, so it is checked against viability.

    Concentration–response% of baseline

    Illustrative model. Responses follow the direction of effects reported for these reference compounds in MEA studies of rodent and human iPSC-derived neurons. Magnitudes and EC₅₀ values are approximate and are not BNP measurements.

    Where current models fall short

    An honest comparison, including what BNP cannot do.

    Every model trades realism against cost and throughput. BNP adds human cells and live network function, and it still needs other models for whole-body exposure and behaviour.

    Criterion2D cell linesRodent studiesIn silico modelsBNP

    Positioning

    BNP adds a human, functional readout earlier in development, so weak candidates fail sooner and at lower cost, and some animal studies can be reduced. It complements animal and clinical data. It does not replace clinical trials.

    Evidence and regulation

    The science and the rules have both moved toward human-based methods.

    Key published work, regulatory changes and industry milestones behind BNP. Filter by type.

    1. 2007
      Science

      Human induced pluripotent stem cells

      Adult human skin cells are reprogrammed into stem cells that can become any cell type, including neurons. This makes patient-specific human neurons possible.4

    2. 2013
      Regulation

      CiPA cardiac safety initiative

      Regulators and industry begin building a proarrhythmia assay that includes human iPSC cardiomyocytes recorded on MEAs. It is the closest precedent for using human-cell electrophysiology in safety decisions.

    3. 2022
      Science

      DishBrain

      Cultured neurons on a high-density MEA learn to play a simplified Pong game in a closed feedback loop.1

    4. Dec 2022
      Regulation

      FDA Modernization Act 2.0

      US law removes the statutory requirement for animal testing before human trials and explicitly allows cell-based assays, organ chips and computer models.14

    5. 2023
      Regulation

      India amends the NDCT Rules

      India’s New Drugs and Clinical Trials Rules are amended to allow non-animal methods such as organoids, organ-on-chip systems and computational models.16

    6. 2023
      Regulation

      OECD neurotoxicity in vitro battery

      OECD publishes recommendations for the developmental neurotoxicity in vitro battery, which includes an MEA-based network formation assay.15

    7. 2023
      Science

      Organoid intelligence framework

      A roadmap for using brain organoids for computation and learning, including the ethics involved.3

    8. Dec 2023
      Science

      Brainoware

      A brain organoid on an MEA is used for reservoir computing and reaches about 78% accuracy on a speaker-identification task.2

    9. Sep 2024
      Regulation

      First organ-chip in FDA’s ISTAND pilot

      The Emulate Liver-Chip is accepted into ISTAND at the letter-of-intent stage, the first step toward qualification for a defined context of use.17

    10. Mar 2025
      Industry

      Cortical Labs launches CL1

      A commercial biological computer built on human neurons grown on an MEA goes on sale, showing that neuron-on-chip hardware can be productised.

    11. Apr 2025
      Regulation

      FDA roadmap to reduce animal testing

      FDA sets out a plan to reduce animal studies in preclinical safety testing, starting with monoclonal antibodies.11

    12. Feb 2026
      Industry

      India landscape report on alternatives

      An industry report finds that unclear acceptance criteria still limit routine use of non-animal data in Indian drug submissions. Validation work is the gap.16

    13. Mar 2026
      Regulation

      FDA draft guidance on NAMs

      FDA’s draft guidance on new approach methodologies sets four validation features: context of use, human biological relevance, technical characterisation and fitness for purpose.12

    14. Apr 2026
      Regulation

      FDA year-one progress report

      ISTAND becomes a permanent programme and FDA creates a database of accepted NAM approaches.13

    What regulators will ask BNP to show

    Mapped to FDA’s four validation features for new approach methodologies.

    Context of use

    One narrow question first

    A NAM is judged for a specific decision, not in general.

    BNP plan: seizure-liability hazard identification for small molecules before first-in-human dosing.

    Human biological relevance

    Human cells that express the targets

    The cells must carry the receptors and channels the drugs act on.

    BNP plan: human iPSC cortical neurons, with target expression and responses to reference compounds documented.

    Technical characterisation

    Reproducible across wells, batches and sites

    Variability between cell batches is the main weakness of iPSC models.

    BNP plan: predefined baseline acceptance criteria, plate-level controls and repeat runs across cell lots.

    Fit for purpose

    Proven against clinical outcomes

    Performance has to be shown on compounds whose human effects are known.

    BNP plan: a blinded reference set, reporting sensitivity and specificity against clinical seizure findings.

    Roadmap

    Build the evidence one stage at a time.

    Each stage produces something useful on its own and de-risks the next.

    Stage 1In progress

    Computational prototype

    Virtual assay and data dashboard, with a feature library calibrated against published MEA pharmacology.

    Stage 2Next

    Partnered wet-lab pilot

    MEA recordings on commercially sourced human iPSC neurons through an academic core facility or CRO, against a reference set of about 20 compounds with known clinical outcomes.

    Stage 3Planned

    Screening service

    Seizure-liability and neurotoxicity screening for pharma, biotech and CROs, reported against a predefined context of use.

    Stage 4Long term

    Patient-derived networks

    Disease-specific iPSC lines, such as genetic epilepsies, for mechanism and efficacy testing, then 3D organoids on high-density arrays.

    Project dashboard

    Open the team’s working dashboard.

    The BNP dashboard is hosted separately and opens in a new tab.

    sg-2512.github.io/bnp-dashboard

    Open dashboard ↗

    Team

    Built at UIPS, Panjab University.

    • Meet Yashraj SinghB.Pharm, UIPS
    • ParasB.Pharm, UIPS
    • SakshiB.Pharm, UIPS

    University Institute of Pharmaceutical Sciences, Panjab University, Chandigarh, India.

    Collaborate

    We are looking for partners.

    • Labs with human iPSC neuron culture or MEA recording capacity
    • Pharma, biotech and CRO teams with CNS compounds to profile
    • Electrophysiology and computational neuroscience mentors

    References

    1. Kagan BJ, Kitchen AC, Tran NT, et al. In vitro neurons learn and exhibit sentience when embodied in a simulated game-world. Neuron. 2022;110(23):3952–3969. doi:10.1016/j.neuron.2022.09.001
    2. Cai H, Ao Z, Tian C, et al. Brain organoid reservoir computing for artificial intelligence. Nature Electronics. 2023;6:1032–1039. doi:10.1038/s41928-023-01069-w
    3. Smirnova L, Caffo BS, Gracias DH, et al. Organoid intelligence (OI): the new frontier in biocomputing and intelligence-in-a-dish. Frontiers in Science. 2023;1:1017235. doi:10.3389/fsci.2023.1017235
    4. Takahashi K, Tanabe K, Ohnuki M, et al. Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell. 2007;131(5):861–872. doi:10.1016/j.cell.2007.11.019
    5. BIO, Informa Pharma Intelligence, QLS Advisors. Clinical Development Success Rates and Contributing Factors 2011–2020. 2021.
    6. Cummings JL, Morstorf T, Zhong K. Alzheimer’s disease drug-development pipeline: few candidates, frequent failures. Alzheimer’s Research & Therapy. 2014;6(4):37. doi:10.1186/alzrt269
    7. Pardridge WM. The blood–brain barrier: bottleneck in brain drug development. NeuroRx. 2005;2(1):3–14. doi:10.1602/neurorx.2.1.3
    8. GBD 2021 Nervous System Disorders Collaborators. Global, regional, and national burden of disorders affecting the nervous system, 1990–2021. Lancet Neurology. 2024;23(4):344–381. doi:10.1016/S1474-4422(24)00038-3
    9. DiMasi JA, Grabowski HG, Hansen RW. Innovation in the pharmaceutical industry: new estimates of R&D costs. Journal of Health Economics. 2016;47:20–33. doi:10.1016/j.jhealeco.2016.01.012
    10. Taylor K, Alvarez LR. An estimate of the number of animals used for scientific purposes worldwide in 2015. ATLA. 2019;47(5–6):196–213. doi:10.1177/0261192919899853
    11. US Food and Drug Administration. Roadmap to Reducing Animal Testing in Preclinical Safety Studies. April 2025.
    12. US Food and Drug Administration. General Considerations for the Use of New Approach Methodologies in Drug Development (draft guidance). March 2026.
    13. US Food and Drug Administration. Reducing Animal Testing in Nonclinical Studies: Year One Progress and the Path Forward. April 2026.
    14. FDA Modernization Act 2.0, enacted in the Consolidated Appropriations Act, 2023 (Pub. L. 117-328). December 2022.
    15. OECD. Initial Recommendations on Evaluation of Data from the Developmental Neurotoxicity (DNT) In-Vitro Testing Battery. Series on Testing and Assessment No. 377. 2023.
    16. Business Today. Why animal testing still drives drug approvals in India, and what it costs pharma. 5 February 2026. businesstoday.in
    17. Emulate. Liver-Chip S1 accepted into FDA’s ISTAND pilot program. 2024. emulatebio.com