Metabolic Insights

Saliva based point-of-care testing technology

Status

In Development

Mii

More to come

While I’m unable to go into specifics of the app’s internals for now, I wanted to note another interesting aspect of the project: the experimental set up.

Context

Metabolic Insights (Mii) was founded with the general goal of empowering clinicians and customers with information about their body, through non-invasive & repeatable techniques. Of particular interest to me is assessing the body’s metabolism, with a focus on identifying intervals and triggers of fat storage versus fat oxidation. Ultimately, the aim is to model the body’s response to specific foods and generate tailored meal plans aligned with individual health goals.

Since the inception of Mii, I have been involved in the development of the technology, which initially focused on creating an electrochemical device capable of detecting multiple analytes in saliva without the need for reagents or consumables typical of a standard lateral flow assay. The project has since evolved to utilize a custom lateral flow assay to detect salivary insulin, a known regulator of metabolism. To support this, I developed a smartphone app that functions as the lateral flow assay reader.

Experimental setup

I find scientific and experimental programming in Swift can be cumbersome, particularly for tasks like plotting/visualization. Python already has great solutions to most data problems, so I created a project structure that leverages Python for analytics, enabling rapid algorithm testing cycles and comprehensive result tracking with MLFlow.

It’s a simple (ie probably common) solution: the core calculation logic is encapsulated within a dedicated, separate framework, allowing it to be wrapped in a Swift executable, which I then import into my Python analysis environment, as well embed directly in the app itself. Obviously, there is some automation that goes on in the background in regards to the build and the import of data but that wasn’t particularly novel to me.

Findings

It’s an obvious setup in retrospect but I had not utilized this project structure before, despite working on similar products. The setup enables me to conduct testing, analysis, and plotting in Python while maintaining consistency with the app’s actual logic. It ensures that the same code is deployed both in the application and during analysis. Much better than Swift playgrounds I find.