Stratum Biosciences, a biotechnology firm leveraging artificial intelligence to accelerate skin biology discovery, has announced the closing of a $2-million seed funding round. The round was led by the Harvard Business School Alumni Angels of Greater New York and Skin Angels, the world’s largest network of early-stage dermatology investors.
The company was established in 2024 by dermatologist Ross Lane Pearlman, MD (CEO), and dermatopathologist Buu Duong, MD (president). The leadership team includes chief scientific officer David Osborne, Ph.D., a former co-founder of Arcutis. Stratum is currently a resident of Johnson & Johnson’s JLABS incubator in SoHo, New York.
As a participant in Nvidia's Inception program, the company utilizes high-performance computing to identify novel biomarkers. Its new capital infusion will be used to scale Stratum’s proprietary SkinSync platform, which analyzes a massive biobank of more than 200,000 real skin samples.
Over the past year, the firm has validated its approach through partnerships with global CPG leaders to develop novel composition-of-matter actives. Initial research targets include high-value intellectual property for skin longevity, hyperpigmentation and eczema.
In this exclusive Q&A, Pearlman discusses the company's technology and what it means for the future of skin care.
Pearlman: Historically, the industry has relied on the bricks and mortar model of the skin barrier—a static view of skin barrier cells. Our SkinSync platform has revealed that the barrier is far more dynamic, fluid, and varied than we previously thought.
Buu Duong, MD, has mentioned that the company is now decoding the skin's biological language. Beyond identifying new ingredients, can this Al-driven understanding of skin's language be used to create personalized bio-resetting routines that adapt to an individual’s real-time biomarker fluctuations?
A compelling example of this is in the field of longevity. We know that skin health and protein expression can be influenced by a range of environmental factors and internal biological processes. By detecting individual and population-based biomarker fluctuations, we can move from generic skin care toward a model of targeted biological context.
So we are not bio-resetting, more like “bio-hacking” the skin barrier with real science to deliver ingredients more effectively without causing barrier breakdown or inflammation. We use our BioBank to find the strategies to do this and then apply the insights to well-characterized, specific skin problems like hyperpigmentation and inflammation.
As for the precision aspect of our approach, that is part of the plan with our current seed funding. Scaling our platform will help us learn more and more from our extensive skin sample BioBank, and archetypes of skin in different conditions and populations will become clearer. We will use these insights to drive more personalized skin care product development and regimen recommendation strategies.
As a member of Nvidia's Inception program, you are leveraging high-performance computing to identify undiscovered targets. Will this technology eventually allow global brands to replace traditional, time-consuming clinical trials with Al-simulated digital twin skin models based on your 200,000-sample biobank? And how much compute is needed to support this type of research and development?
Pearlman: Our current use case is focused on discovering new targets by studying the protein structure of skin in different conditions and populations at higher-resolution and scale than ever before. Our partners will be the first to discover and understand how to access these critical targets.
The SkinSync platform is specifically built for this workflow: first, study the condition and representative population in our Discover module using our BioBank data. Then, we integrate those insights into design of small molecules or peptides in our Design module and optimize bioactivity, tolerability, and barrier permeability. Finally, the lead candidates are simulated with thousands of different formulations in our Deliver module to find the best formulation options to deliver the active ingredient.
In the future, we will evolve these models to build machine-learning/AI algorithms for in silico or “digital twin” simulations. This technology will move skin care clinical trials from the end of the R&D cycle to the beginning. While we don't see traditional trials disappearing, our participation in the NVIDIA Inception program allows us to build high-fidelity permeation models of skin that can simulate active ingredients in thousands of formulations. This allows our development partners to “fail fast” in a digital environment, ensuring that only the most promising ingredient-formulation pairings ever reach a clinical trial.
As for the compute, it is a very heavy burden. We are in the process of expanding our capabilities by building completely customized multi-GPU units with massive associated storage capacities to parse and understand these data (think >1 petabyte of data).
