The Galectin Intelligence Platform.

From Human Pathway Perturbation to Clinical Precision

Built on real-world patient biomarker and clinical data from Galectin-3 modulation, the platform converts every trial readout into sharper predictive models, leaner clinical designs, and new therapeutic targets.

Galectin Intelligence Platform

The TB006 clinical AI flywheel

TrueBinding’s goal is to use AI to find patterns in TB006 study data. Each new trial can test whether those patterns hold up.

Step 01

Study TB006

Clinical studies investigate TB006, an antibody that targets Galectin-3.

What does a study measure?

The study protocol defines which outcomes researchers will assess and when. Those choices determine which questions the study data can help answer.

Step 02

Collect human data

Researchers track biomarkers and clinical assessments over time.

Why measure more than once?

A measurement taken before treatment provides a starting point. Later measurements show how that participant’s biomarker levels or clinical assessments changed during the study.

Step 03

Explore patterns with AI

Researchers can explore how starting measurements relate to later outcomes across participants.

What might that look like?

For example, do participants who start with similar biomarker levels also show similar changes over time? A pattern seen in one study gives researchers a question to test in the next.

Step 04

Test the pattern in the next trial

A new group of participants gives researchers a way to examine whether an earlier finding holds up.

What closes the loop?

If a pattern appears again, the team can investigate it further. If it changes or disappears, that result can help researchers revise the question they ask in the next trial.

Each trial makes the next one smarter.

Galectin Intelligence Platform

The TB006 clinical AI flywheel

TrueBinding’s goal is to use AI to find patterns in TB006 study data. Each new trial can test whether those patterns hold up.

Step 01

Study TB006

Clinical studies investigate TB006, an antibody that targets Galectin-3.

What does a study measure?

The study protocol defines which outcomes researchers will assess and when. Those choices determine which questions the study data can help answer.

Step 02

Collect human data

Researchers track biomarkers and clinical assessments over time.

Why measure more than once?

A measurement taken before treatment provides a starting point. Later measurements show how that participant’s biomarker levels or clinical assessments changed during the study.

Step 03

Explore patterns with AI

Researchers can explore how starting measurements relate to later outcomes across participants.

What might a model look for?

For example, do participants with similar starting biomarker levels also show similar changes on a clinical assessment? A pattern in one study gives researchers a specific question to test again.

Step 04

Test findings in the next study

A new group of participants gives researchers a way to examine whether an earlier finding holds up.

What closes the loop?

If a pattern appears again, the team can investigate it further. If it changes or disappears, that result can help researchers revise the question they ask in the next trial.

Each trial makes the next one smarter.

Galectin Intelligence Platform

The TB006 clinical AI flywheel

TrueBinding’s goal is to use AI to find patterns in TB006 study data. Each new trial can test whether those patterns hold up.

Step 01

Study TB006

Clinical studies investigate TB006, an antibody that targets Galectin-3.

What does a study measure?

The study protocol defines which outcomes researchers will assess and when. Those choices determine which questions the study data can help answer.

Step 02

Collect human data

Researchers track biomarkers and clinical assessments over time.

Why measure more than once?

A measurement taken before treatment provides a starting point. Later measurements show how that participant’s biomarker levels or clinical assessments changed during the study.

Step 03

Explore patterns with AI

Researchers can explore how starting measurements relate to later outcomes across participants.

What might a model look for?

For example, do participants with similar starting biomarker levels also show similar changes on a clinical assessment? A pattern in one study gives researchers a specific question to test again.

Step 04

Test findings in the next study

A new group of participants gives researchers a way to examine whether an earlier finding holds up.

What closes the loop?

If a pattern appears again, the team can investigate it further. If it changes or disappears, that result can help researchers revise the question they ask in the next trial.

Each trial makes the next one smarter.

Galectin Intelligence Platform

The TB006 clinical AI flywheel

TrueBinding’s goal is to use AI to find patterns in TB006 study data. Each new trial can test whether those patterns hold up.

Each trial makes the next one smarter.

  1. Study TB006

    Clinical studies investigate TB006, an antibody that targets Galectin-3.

  2. Collect human data

    Researchers track biomarkers and clinical assessments over time.

  3. Explore patterns with AI

    Can measurements taken at the start of a study help explain later outcomes?

  4. Test findings in the next study

    Does a pattern seen in one study hold up in a new group of participants?

Platform Architecture & Clinical Applications

How Human Perturbation Data Drives Every Trial

Rather than training on synthetic assumptions, our computational stack ingests multi-dimensional clinical and biomarker readouts from TB006-treated patients to sharpen trial execution and antibody discovery.

End-to-End Computational Architecture

01 / HUMAN DATA INPUTS

Multi-Modal Clinical Readouts

Longitudinal plasma Galectin-3 kinetics (~9x target engagement), PK/PD profiles, inflammatory biomarker panels, and clinical cognitive/motor endpoints across neurodegenerative cohorts.

→
02 / MODELING LAYER

Trajectory & Subgroup Analytics

Baseline covariate modeling, untreated disease-trajectory forecasting, and multi-dimensional responder clustering built on real human pathway perturbation.

→
03 / CLINICAL OUTPUTS

Trial & Pipeline Execution

Prognostic covariate adjustment for leaner Phase 2/3 trials, biomarker-guided inclusion criteria, and epitope prioritization for next-generation antibodies.

Trial Sensitivity & Power

Prognostic digital twins

For each trial participant, the platform models the disease trajectory they would most likely follow without treatment based on baseline clinical and biomarker covariates. Incorporating these prognostic scores reduces unexplained outcome variance—enabling smaller sample sizes and higher statistical confidence under regulatory-qualified prognostic frameworks.

Pipeline Application: Informing covariate-adjusted analysis and trial design across our 210-patient Phase 2b Alzheimer’s study (TB006-AD-2105) and future pivotal programs.
Precision Stratification

Responder enrichment

Neuroinflammatory and neurodegenerative patients do not respond uniformly. By mining longitudinal human datasets from completed Phase 1b/2a studies and ongoing Phase 2 cohorts, the platform identifies baseline biomarker and clinical signatures linked to stronger Gal-3 target engagement and clinical benefit—preventing signal dilution in subsequent trials.

Pipeline Application: Translating multi-indication human response patterns across Alzheimer’s and Parkinson’s trials into sharper patient-selection criteria.
Next-Gen Antibodies

Human-first target discovery

Pharmacological neutralization of Galectin-3 in people provides a direct, in vivo readout of how the pathway regulates chronic inflammation and protein aggregation in human biology. The platform analyzes downstream biomarker shifts to guide affinity and epitope engineering for next-generation Gal-3 antibodies and uncover adjacent pathway targets.

Pipeline Application: Extending insights from systemic and CNS Gal-3 modulation into follow-on biologics and new therapeutic indications.

What can researchers learn from a TB006 study?

Explore how measurements collected during a study could help researchers identify a pattern and decide what to test next.

Choose a stage to explore

Measure change over time

A measurement taken before treatment gives researchers a starting point. Later visits let them compare changes in biomarkers with changes in clinical assessments.

Illustration of measurements across four visits

Look for patterns across participants

Do participants with similar starting biomarker levels also show similar changes later? A model could help researchers explore that question across a study population.

Illustrative groups formed from example observations

Choose a question for the next study

If a pattern appears, researchers can specify what to measure in a new group of participants and test whether the finding appears again.

A study observation leads to a new research question
Prognostic digital twins

Could measurements taken at enrollment help estimate a participant’s likely disease course without treatment? Researchers could compare those estimates with observed outcomes to assess how well the model performs.

Differences in response

Do starting characteristics help explain why participants have different outcomes? A pattern found in one group would give researchers a hypothesis to examine in another.

Galectin-3 research

Which biomarker changes warrant a closer look after a TB006 study? Those observations could help researchers prioritize follow-up questions about Galectin-3 biology.

Scientific Discipline & Guardrails

What the Platform Is Not

TrueBinding treats artificial intelligence as an instrument for rigorous biology—never a substitute for experimental or clinical proof.

01 / Wet-Lab First

Not a replacement for wet-lab validation

Computational predictions generate hypotheses, not proof. No AI-derived antibody candidate or target insight advances without rigorous biochemical, cellular, and in vivo validation.

02 / Evidence-Earned

Not a promise beyond the data

We do not treat algorithms as black-box oracles or extrapolate past what human biology supports. Every prognostic model and responder signature must be earned through reproducible clinical readouts.

03 / Measured Impact

Not unmeasured efficiency claims

We do not publish hypothetical timeline cuts or speculative cost-reduction projections. Operational and trial-design gains are reported only after they are empirically measured in prospective studies.

Our Mandate: The platform’s job is to make the science faster and the trials sharper—strictly on evidence.

How we evaluate AI findings

An AI model can suggest a question worth testing. The evidence we need depends on what the model predicts.

Test biological predictions in the lab

When a model points to an antibody candidate or a potential target, the prediction is a starting hypothesis. Before it advances, researchers test it in biochemical, cellular, and in vivo studies.

Check whether clinical patterns repeat

A pattern in one group of participants raises a question for the next study. Does the same relationship appear again? Reproducible clinical observations help researchers judge how much confidence to place in a model.

Measure the impact on trial design

A proposed change to a trial may look promising on paper. TrueBinding reports gains in trial design or operations only after measuring them in prospective studies.

Founder & Leadership

George Hong

Founder & Chief Executive Officer

Education M.S. in Computer Science, University of Arizona
AI Infrastructure Leadership CEO, Wangsu Technology (SZSE: 300017) — leading its pivot into AI infrastructure for AI models
TrueBinding Founded in 2016
Healthcare Investment Portfolio
Ceribell Neurovasc Arrinex Devoro Lucira
Data-Native Leadership

Why a Computer Scientist Founded a Biotech

“Drug development’s bottleneck is no longer chemistry — it is knowing which patients, which trials, and which targets. The company is built the way a data company is built: the drug generates the data, and the data improves the drug.”

George Hong founded TrueBinding in 2016 on that simple observation. That closed-loop architecture is the Galectin Intelligence Platform, and it is the reason a computer scientist built a clinical-stage biotechnology company.

Alongside leading TrueBinding, he serves as Chief Executive Officer of Wangsu Technology (SZSE: 300017), where he is spearheading the company’s strategic transformation into large-scale AI infrastructure for AI models—bringing deep systems-engineering and data-platform discipline directly into therapeutic development.

Founder & Leadership

George Hong

George Hong

Founder & Chief Executive Officer

Education
M.S. in Computer Science, University of Arizona
AI Infrastructure Leadership
CEO, Wangsu Technology (SZSE: 300017) — leading its pivot into AI infrastructure for AI models
TrueBinding
Founded in 2016

Healthcare Investment Portfolio

  • Ceribell
  • Neurovasc
  • Arrinex
  • Devoro
  • Lucira

Data-Native Leadership

Why a Computer Scientist Founded a Biotech

“Drug development’s bottleneck is no longer chemistry — it is knowing which patients, which trials, and which targets. The company is built the way a data company is built: the drug generates the data, and the data improves the drug.”

George Hong founded TrueBinding in 2016 on that simple observation. That closed-loop architecture is the Galectin Intelligence Platform, and it is the reason a computer scientist built a clinical-stage biotechnology company.

Alongside leading TrueBinding, he serves as Chief Executive Officer of Wangsu Technology (SZSE: 300017), where he is spearheading the company’s strategic transformation into large-scale AI infrastructure for AI models—bringing deep systems-engineering and data-platform discipline directly into therapeutic development.

George Hong

Founder & Leadership

George Hong

Founder & Chief Executive Officer

Education
M.S. in Computer Science, University of Arizona
AI infrastructure leadership
CEO, Wangsu Technology (SZSE: 300017)
Healthcare investment portfolio

Ceribell, Neurovasc, Arrinex, Devoro, and Lucira

Why a Computer Scientist Founded a Biotech

George Hong founded TrueBinding in 2016. His background in computer science informs the idea behind the Galectin Intelligence Platform: what researchers learn in one clinical study can shape the questions they ask in the next.

“The company is built the way a data company is built: the drug generates the data, and the data improves the drug.”

George Hong

George Hong

Founder & Chief Executive Officer

Education
M.S. in Computer Science, University of Arizona
AI infrastructure leadership
CEO, Wangsu Technology (SZSE: 300017)
Healthcare investment portfolio

Ceribell, Neurovasc, Arrinex, Devoro, and Lucira

Why a computer scientist founded a biotech

George Hong founded TrueBinding in 2016. His background in computer science informs the idea behind the Galectin Intelligence Platform: what researchers learn in one clinical study can shape the questions they ask in the next.

“The company is built the way a data company is built: the drug generates the data, and the data improves the drug.”

George Hong
Founder & Chiever Executive Officer

Founder perspective

Why a Computer Scientist Founded a Biotech

George Hong founded TrueBinding in 2016. His background in computer science informs the idea behind the Galectin Intelligence Platform: what researchers learn in one clinical study can shape the questions they ask in the next.

“The company is built the way a data company is built: the drug generates the data, and the data improves the drug.”

— George Hong
Leadership background and investments
Education
M.S. in Computer Science, University of Arizona
AI infrastructure leadership
Chief Executive Officer, Wangsu Technology (SZSE: 300017)
Healthcare investment portfolio
Ceribell, Neurovasc, Arrinex, Devoro, and Lucira