Contents

Pages

Contact Griffin Labs

Previous innovation Applied AI 5 min read

Self-updating, multiscale simulations of human biology.

Highlights

  1. Human biology research is fragmented and overwhelming: 75,000+ intervention studies for Alzheimer’s alone.
  2. SUMMIT’s domain-expert AI agents unify thousands of causal connections from the literature in minutes rather than weeks.
  3. Researchers can simulate interventions across molecular, cellular, organ and behavioral scales and identify novel combination therapies.

1 The problem

A fragmented and overwhelming research ecosystem in human biology.

1 Exploding research volume
75,000+ intervention studies for Alzheimer’s alone, and hundreds of new articles published every week.
2 Multiscale complexity
Human biology spans molecular → cellular → organ → behavioral levels.
3 Integration bottleneck
Capturing cross-scale interactions from thousands of papers is incredibly laborious if done by hand.
4 Ephemeral insights
By the time findings are integrated, new studies emerge, deprecating much of the previous effort.

2 Our solution

SUMMIT unifies the literature in a causal graph.

SUMMIT, the Self-Updating Multiscale Modeling and Integration Toolkit, is the world’s first self-updating, multiscale platform for biological simulations. It unifies thousands of causal connections from the scientific literature in minutes rather than weeks.

Existing alternatives are either limited to a single biological scale or require manual updates. SUMMIT is the first to combine both capabilities, so researchers can simulate therapies and visualize complex network effects across a vast biological landscape.

Because SUMMIT integrates the latest scientific findings at the push of a button, researchers can identify novel combination therapies and explore biological interactions with unprecedented speed and scope.

Figure 1

How SUMMIT works

  1. aInput

    Any biological variable: a disease, target, drug, supplement or protocol.

    • Alzheimer’s disease
    • Intermittent fasting
    • Sleep quality
    • Resveratrol
  2. bExtract

    An ensemble of domain-expert AI agents extracts interactions reported in the literature and unifies them in a causal graph.

  3. cExplore

    Identify therapeutic targets, or simulate interventions and see network effects.

Figure 1. From a single variable to a causal graph unified from the scientific literature, ready for target identification and simulation.

3 Features

Features

Six short recordings of SUMMIT, from building a causal graph to simulating an intervention.

Figure 2

SUMMIT in use

a Create a causal graph for any disease, target, drug, supplement or protocol

b Expand the graph based on interest

c Each interaction is extracted from a peer-reviewed article

d Scales spanning molecular, cellular, organ and behavioral levels

e Find novel ways to optimize a variable of interest

f Simulate interventions and see network effects

Figure 2. Screen recordings of the SUMMIT interface. Each clip plays without sound when it scrolls into view.

4 The team

The team

Elias Schlie on LinkedIn

AI engineer

As lead software developer, implemented SUMMIT’s agentic workflow and user interface, optimizing latency and scalability.

Open-source contributor on AI for science and agentic workflows; AI BSc summa cum laude in two years.

Andrei Oprea

UX engineer

Contact

Request a demo.

Write to us to see SUMMIT in action, or to ask about the project.