Why is food processing of interest to Schrödinger?

Schrödinger has a long history of applying physics-based modeling to accelerate research and development in pharmaceutical and materials discovery. In my mind, food science is a blend of both life and materials science problems at scales that range from single molecules to the whole food products you find on a shelf in your local grocery store. By bridging that gap, we can help researchers make more informed decisions faster when developing new, or refining existing food processing methods. Combining these scientific advances with the advances in computing power over the past 15 years, simulations of relevant food products are now tractable.

How does modeling help companies improve or advance?

Our solutions can predict properties for both single molecules and full food products, and can provide novel insight at molecular level in real time. Experimental methods to characterize food processing conditions often lack the resolution to rationalize physicochemical behavior, and often reduce a complex process down to a single set of numbers. By using a bottom up approach, researchers can understand how their products are behaving from a molecular level. This can help reduce the number of experiments and accelerate the time to generate meaningful data.

How can machine learning help processing be more sustainable?

There are many parameters in process modeling, and often they require data collection at the pilot or product scale level. This data can be used to build machine learned models to optimize processes, and can cover everything from ingredient chemistry and composition to rheological properties. By incorporating chemical information into your model, you are able to add many more parameters that not only increase the accuracy of a ML model, but also the interpretability. By reducing the number of experiments and pilot studies, models are inherently sustainable themselves. Moreover, by optimizing processes, whether it is chemistry based or larger scale, can help reduce water consumption, heat generation, and packaging materials for food products.

What’s next in innovative technologies for the food industry? And what is Schrödinger’s role in this advancement?

As we face changes to established agricultural practices due to climate change, combined with volatile markets affecting research and development budgets, the need for streamlined processes and models will become crucial to deliver innovative food products. While most food companies have thus far been slow in adopting digital transformation and predictive modeling due to either lack of understanding or resistance to change established practices, this will soon become unsustainable. In order to win in competitive markets, while also producing sustainable food raw materials, products, and packaging, companies will need to rely more on digital methods to deliver cost effective solutions. Digitization and machine learning are hot topics today as more companies push to apply these tools to food product development to stay ahead of the curve. I see the incorporation of physics-based simulation and ML models, whether they are used in product development, as digital twins, or to gain deeper scientific understanding as being a necessary step forward in this evolution of the industry. Schrödinger technology can play an integral role in shifting today’s approach from reactive use of modeling (i.e. in instances where a product is not behaving as expected) to predictive modeling that guides new product development. We envision customers using Schrödinger’s software throughout the development process, especially at early stages to drive innovations.

Join Jeff at Summit Americas April 26-27 for more information at his presentation “Innovating plant-based food using a digital chemistry strategy.” More details and register here.