Insitro Integrated ML Models Into Lilly TuneLab
The machine learning models aim to improve drug discovery by predicting small molecule behavior.
Updated on Oct. 6, 2026 in Biotech
Live Poll
Do you believe shared research data helps pharmaceutical companies develop new treatments more effectively?
Insitro has successfully integrated its proprietary machine learning models into the Eli Lilly and Company TuneLab platform. These models are designed to predict both in vitro and in vivo properties of small molecules using extensive preclinical data.
Why it matters
By refining design-stage predictions, these models aim to prevent the advancement of flawed drug candidates and reduce the industry's reliance on animal studies. This capability helps streamline the drug development pipeline, which often faces high failure rates for programs entering clinical trials.
The models are trained on multi-species preclinical data gathered from hundreds of thousands of unique molecules over decades of research. The platform utilizes a federated learning infrastructure to process these properties.
The players
Insitro
This South San Francisco-based company focuses on using machine learning to accelerate drug discovery and development.
Eli Lilly and Company
This major pharmaceutical corporation provides the TuneLab platform to offer biotech companies access to advanced drug development tools.
The details
Insitro developed these models to analyze small molecule behavior with greater precision than traditional methods. The technology leverages decades of research to filter candidates before they reach the cost-intensive stages of clinical testing.
Timeline
2018: Insitro began building its multi-modal data corpus.
September 2025: Insitro and Lilly announced their partnership.
April 2025: FDA released a roadmap for reducing animal testing.
March 2026: FDA issued draft guidance on new approach methodologies.
October 6, 2026: Insitro announced its models are available in Lilly TuneLab.
The Tech Race
This integration follows the regulatory trajectory established by the FDA's 2025 roadmap for reducing animal testing by providing a computational alternative to preclinical animal studies. It represents a larger industry shift toward replacing traditional, time-intensive laboratory methods with predictive digital workflows.
While this technology functions behind the scenes in laboratory settings, its success could lead to safer, more effective pharmaceutical products reaching patients sooner. By reducing preclinical failure rates, these tools help optimize the focus of resources toward viable treatments.
The takeaway
The implementation of federated learning in drug discovery signifies a move toward more data-efficient pharmaceutical research. Developers and biotech firms can utilize these models to catch potential issues in molecule design significantly earlier in the research cycle.
What happens next
Insitro plans to enter nine first-in-class programs into clinical trials in early 2027.
Further reading
Learn more about the latest industry developments in our Biotech section.
Source note: This article includes information reported by Firstwordpharma.
Live Poll
Do you believe shared research data helps pharmaceutical companies develop new treatments more effectively?
